MMI 403: Introduction To Medical Informatics …medicalinformatics.weebly.com › uploads › 2 ›...
Transcript of MMI 403: Introduction To Medical Informatics …medicalinformatics.weebly.com › uploads › 2 ›...
MMI 403: Introduction To Medical Informatics
Biomedical Informatics: Computer Applications in Health Care and Biomedicine Series: Health Informatics Shortliffe, Edward H. (Ed.) 3rd ed., 2006 Springer Science+Business Media, LLC, XXVI, 1037 p., 159 illus., 4 in color, Hardcover ISBN: 978-0-387-28986-1
Contents Chapter 1: The Computer Meets Medicine and Biology: Emergence of a Discipline ................................... 9
Record system ....................................................................................................................................... 9
4 major issues for EHR: ......................................................................................................................... 9
To pool and integrate data, need: ...................................................................................................... 10
- Creation of a National Health Information Infrastructure (NHI), ............................................... 10
Cyclical Information Flow: ................................................................................................................... 11
Requirements for achieving the vision ............................................................................................... 11
The use of Computers in Biomedicine ................................................................................................ 11
Work in bioinformatics ....................................................................................................................... 12
Chapter 2: Biomedical Data: Their Acquisition, Storage, and Use .............................................................. 13
A patient data can be defined as: ....................................................................................................... 13
Types of Medical data: ........................................................................................................................ 13
Uses of Medical Data .......................................................................................................................... 14
Weaknesses of the traditional medical record system: ...................................................................... 15
Strategies of Medical data selection and use ..................................................................................... 16
The Hypothetico deductive approach ................................................................................................. 17
Chapter 19: Information Retrieval and Digital Libraries ............................................................................. 18
- Evolution ..................................................................................................................................... 18
Biomedical Information: ..................................................................................................................... 19
4 states of information need .............................................................................................................. 19
Criteria for a webpage ........................................................................................................................ 20
Process of EBM: .................................................................................................................................. 21
4 major foreground questions categories: ......................................................................................... 21
The 4S model of hierarchy of EBM resources: .................................................................................... 21
Classify content categories: ................................................................................................................ 21
Indexing ................................................................................................................................................... 23
3 Categories: ....................................................................................................................................... 24
3 components of UMLS knowledge source ........................................................................................ 24
Word Indexing Limitations: ................................................................................................................. 25
Retrieval .................................................................................................................................................. 26
Retrieval Systems .................................................................................................................................... 26
Evaluation of IR system ........................................................................................................................... 27
Chapter 6: System Design and Engineering in Health Care ........................................................................ 29
Users ................................................................................................................................................... 29
Functions of a Computer System ............................................................................................................ 30
Need cost-benefit and cost effectiveness analysis ............................................................................. 31
Why Fail? ................................................................................................................................................. 32
How to Design For Success ..................................................................................................................... 32
A data flow diagram (DFD) ...................................................................................................................... 33
Data Storage............................................................................................................................................ 33
Water Fall model for development process ........................................................................................... 33
Alternative Methodologies ................................................................................................................. 34
Parameters to consider For Successful System Implementation: ...................................................... 35
Chapter 5: Essential concept for Biomedical Computing ........................................................................... 35
Computer architecture, e.g. a simple Von Newman machine ............................................................ 36
Memory ............................................................................................................................................... 36
3 conventions for composing internet names from segments ........................................................... 38
Data Management .................................................................................................................................. 39
Database Management System .............................................................................................................. 39
4 levels stack of TCP/IP ....................................................................................................................... 40
2 Parameters determined how closely the digital data represent the original analog signal ............ 41
Data and system Security ........................................................................................................................ 41
The security steps, serve 5 key functions: .......................................................................................... 41
Chapter 12: Electronic Health Record System ............................................................................................ 42
Purpose of a patient record is to ........................................................................................................ 42
Benefits of EHR.................................................................................................................................... 42
Functional Components of EHR .......................................................................................................... 43
Issues for Computer based record system.......................................................................................... 44
Data Display ........................................................................................................................................ 45
Query and Surveillance System .......................................................................................................... 46
Challenges Ahead for EHR ................................................................................................................... 46
Chapter 7: Standards in Biomedical informatics......................................................................................... 47
Standard Development Process .......................................................................................................... 47
Standards Development Organizations .............................................................................................. 48
Data Interchange Standards ................................................................................................................... 51
Data Interchange Standards ............................................................................................................... 51
Directions For Standards ..................................................................................................................... 53
Chapter 10: Ethics and Health Informatics: Users, Standards, and Outcomes........................................... 54
Appropriate Users and Educational Standards ................................................................................... 54
10 Criteria for systems scrutiny .......................................................................................................... 55
3 reasons to use clinical computer programs to guide policy ............................................................ 56
Clinical Software applications can be viewed as a product or a service ............................................. 57
Recommendations .............................................................................................................................. 57
Principles for Decision making ............................................................................................................ 57
Chapter 3: Biomedical Decision Making Probabilistic Clinical Reasoning .................................................. 59
Diagnostic Process Stages ................................................................................................................... 59
Cognitive heuristics ............................................................................................................................. 60
Measure of Test performances ........................................................................................................... 62
Index test performed in 2 groups ....................................................................................................... 63
Because of ........................................................................................................................................... 64
*** To Rule in a Diagnosis, Choose a test with high Specificity or a high LR+ *** ............................. 67
*** To Rule Out a disease, choose a test with a high sensitivity or a high LR- *** ............................ 67
Cautions using Bayes Theorem ........................................................................................................... 67
Expected Value Decision Making ........................................................................................................ 68
To use expected value decision making .............................................................................................. 69
4 steps in decision analysis ..................................................................................................................... 69
Represent patient’s preferences with utilities .................................................................................... 70
Decision whether to treat, test, or do nothing ................................................................................... 70
Decision models .................................................................................................................................. 71
Chapter 16: Patient Care System (1 slide) .................................................................................................. 71
Complicating Factors for caregivers .................................................................................................... 71
Information to support patient care ................................................................................................... 72
3 information categories: .................................................................................................................... 72
4 types of Information Processes........................................................................................................ 72
Today’s patient care system ............................................................................................................... 72
Societal forces have influenced the decision and implementation of patient care system ............... 73
Patient Care Systems: ......................................................................................................................... 73
System Uses: ....................................................................................................................................... 74
New systems need to .......................................................................................................................... 74
Data Aggregate to: .............................................................................................................................. 74
Best Patient care Systems: .................................................................................................................. 74
Cornerstone of good patient care system: ......................................................................................... 75
4 Categories: Typology of Science in Medical Informatics ................................................................. 75
5 key factors for successful implementation ...................................................................................... 76
Research to: ........................................................................................................................................ 76
Chapter 15: Public Health Informatics and The Health Information Infrastructure ................................... 77
Public Health Focus ............................................................................................................................. 77
The nature of a given intervention is determined by ......................................................................... 77
Public health 3 core functions: ........................................................................................................... 77
10 essential services of Public Health ................................................................................................. 78
Public health information systems features: ...................................................................................... 78
Public health informatics example ...................................................................................................... 79
Challenges in 4 areas........................................................................................................................... 80
Most of the community Health information networks are not successful because .......................... 80
Benefits of NHII ................................................................................................................................... 81
Barriers and challenges to NHII ........................................................................................................... 81
Approaches to accelerating NHII process ........................................................................................... 82
Vision for application of NHII to homeland security ........................................................................... 83
Challenge of HII in homeland security ................................................................................................ 83
Chapter 13: Management of Information in HC organization .................................................................... 83
Forms Integrated Delivery Networks (IDNs) ....................................................................................... 84
Goals of IDNs ....................................................................................................................................... 84
HCO Operational Information needs .................................................................................................. 85
HCO Integration need ......................................................................................................................... 85
HCO must adapt a 3-pronged approach to securing information ...................................................... 86
Benefits of HCIS ................................................................................................................................... 86
Managing IS in a Changing HC environment ....................................................................................... 87
Functions and components of a HCIS ................................................................................................. 88
Contract Management Systems .......................................................................................................... 89
Historical evolution of HIS ................................................................................................................... 89
Lessons learned: .................................................................................................................................. 90
Forces that will shape the future of HIS.............................................................................................. 92
Chapter 14: Consumer Health Informatics and Telehealth ........................................................................ 92
Historical Example ............................................................................................................................... 93
Engaging consumers in Health Care.................................................................................................... 94
Categories of telehealth and consumer health informatics ............................................................... 94
CHI resources provide patients with condition specific and disease specific information about the
problems they face ............................................................................................................................. 94
Remote monitoring focuses on management, rather than on diagnosis ........................................... 95
Challenges and Future directions ....................................................................................................... 96
Use 6 criteria to evaluate .................................................................................................................... 97
Chapter 4: Cognitive Science and Biomedical Informatics ......................................................................... 98
Cognitive Science can provide insight into ......................................................................................... 98
3 Areas: ............................................................................................................................................... 99
Cognitive research, theories, and methods can contribute to applications in informatics in ............ 99
Logical positivism: All statements are either ...................................................................................... 99
Protocol analysis: The most common method of data analysis ......................................................... 99
E.g. 2 interdependent principles by which we can characterize cognitive systems are: .................. 100
A cognitive layer with 2 long term memory modules ....................................................................... 100
Human memory: ............................................................................................................................... 100
E.g. An epistemological framework of .............................................................................................. 102
Experts worked forward ................................................................................................................... 103
Different levels of expertise .............................................................................................................. 103
Hypothetico-deductive process ........................................................................................................ 104
Usability methods: ............................................................................................................................ 105
Usability includes 5 attributes: ......................................................................................................... 105
- Heuristic Evaluation .................................................................................................................. 105
- Cognitive Engineering ............................................................................................................... 106
Norman’s 7 stage model of action .................................................................................................... 106
2 primary points for break down ...................................................................................................... 106
5 major classes of representation types: .......................................................................................... 107
Criteria for evaluating the efficacy of a representation ................................................................... 108
6 Ways external representations can amplify cognition: ................................................................. 108
DC has 2 central points of inquiry: .................................................................................................... 108
Mediating Role of Technology: ......................................................................................................... 109
The use of clinical practice guidelines desired results ...................................................................... 110
Guideline Development Process ....................................................................................................... 110
Process of translation from Internal Representations into natural and computer represent able
languages .......................................................................................................................................... 111
Chapter 9: Imaging and Structural Informatics ......................................................................................... 112
Imaging Parameters .......................................................................................................................... 113
Perfect Image Modality ..................................................................................................................... 113
4 areas of development .................................................................................................................... 113
4 Image Procession Steps .................................................................................................................. 114
Mathematical models use to aid in the performance of image analysis subtasks ........................... 115
Image Processing Techniques Example ............................................................................................ 115
3D Image Processing ......................................................................................................................... 115
Volume Registration ......................................................................................................................... 116
Functional Imaging ............................................................................................................................ 118
Chapter 11: Evaluation and Technology Assessment ............................................................................... 119
5 Reasons to study clinical Information resources ........................................................................... 119
Stakeholders in Evaluation Studies ................................................................................................... 119
Challenges of study design and conduct ........................................................................................... 119
Full Range of what can be studied .................................................................................................... 120
Evaluating focus changes in different aspects .................................................................................. 121
Factors that characterize an evaluation study .................................................................................. 121
Anatomy of all studies ...................................................................................................................... 121
Major Premises underlying the objectivist approach: Logical Positivist........................................... 121
Intuitionist Pluralist Subjectivist approaches to evaluation ............................................................. 122
Approaches to Evaluation ................................................................................................................. 122
Stages of technology Assessment ..................................................................................................... 123
Stages of Evaluations ........................................................................................................................ 123
Types of studies ................................................................................................................................ 125
4 possible outcomes ......................................................................................................................... 126
Threats to Internal Validity ............................................................................................................... 126
Cost effectiveness and Cost benefit studies ..................................................................................... 127
Determine whether the benefit ($) is larger than the cost ($) ......................................................... 127
Methodology of formal Systems analysis ......................................................................................... 128
Differences between objectivist and subjectivist approaches ......................................................... 128
Natural history of a subjectivist study .............................................................................................. 129
Data Collection and Data Analysis Methods ..................................................................................... 129
The Mindset of evaluation and technology assessment .................................................................. 130
Chapter 20: Clinical Decision Support System .......................................................................................... 131
Types of Decisions ............................................................................................................................. 131
Requirements for decision making ................................................................................................... 131
3 types of decision support functions ............................................................................................... 131
Leeds Abdominal Pain System .......................................................................................................... 132
Change in acceptance of CSS due to: ................................................................................................ 132
Characterize CDSS by: ....................................................................................................................... 132
Development of a knowledge based system: ................................................................................... 133
Barriers to CDSS: ............................................................................................................................... 134
Examples of CDSS .............................................................................................................................. 135
Patient Management: Guideline Based Architectures and the EON system .................................... 135
Several tasks that can benefit from automation .............................................................................. 135
Major components of EON ............................................................................................................... 135
EON shows ........................................................................................................................................ 136
Chapter 21: Computers in Medical Education .......................................................................................... 136
Goals of Medical Education .............................................................................................................. 137
Advantages of using computers in medical education ..................................................................... 137
Computers Learning Methods: ......................................................................................................... 138
Current Applications: ........................................................................................................................ 139
Design, Development, and Technology: Design of Computer based learning applications ............. 139
Application development .................................................................................................................. 140
Technology Considerations ............................................................................................................... 141
Evaluation ......................................................................................................................................... 141
Chapter 24: The Future of computer applications in Biomedicine ........................................................... 141
Major challenges and opportunities facing the field of computational biology .............................. 142
Computers integrated into medical settings by................................................................................ 142
Potential costs of using computers ................................................................................................... 143
Factors affecting the role of computers in medicine ........................................................................ 143
Chapter 1: The Computer Meets Medicine and Biology: Emergence of a
Discipline - Page 1.4
- At the heart of the evolving clinical workstation lies the medical record in a new incarnation:
1) Electronic
2) Accessible
3) Confidential
4) Secure
5) Acceptable to clinicians and patients
6) Integrated with other types of non-patient specific info
- Analyze the processes associated with the creation and use of medical records
Record system
o Integration of processes for data capture and for merging information from diverse
sources
o Easy to access and display needed data
o Easy to analyze
o Share with others
- Page 1.6
- EMR: Best viewed as a set of processes that an organization must put into place, supported by
technology
- A system integration task
- The importance of EMR to support clinical trials: Experiments in which data from specific
patients interactions are pooled and analyzed
- Eliminate the manual task of extracting data from charts
- Ensure compliance with research protocol
- Ensure data elements needed for the trial are compatible with the local EMR conventions
- Page 1.8
4 major issues for EHR:
1) The need for standards in the area of clinical terminology
2) Concerns regarding data privacy, confidentiality and security
3) Challenges of data entry by MD
4) Integration with other information sources
- Page 1.9
- MD seeks broad functionality across a wide variety of systems and resources
- System needs to offer a critical mass of functionality
- Page 1.10
- Clinical data repository: a central computer that gathers and integrates clinical data from diverse
sources
- The increasing investment in the creation of clinical guidelines and pathways
- To reduce practice variability and to develop consensus approaches to recurring management
problems
- We need better methods for delivering the decision to the point of care
- Integrate decision support tools with EHR
- Page 1.13
- Disease surveillance, prevention, and care can be influenced by IT
- Data from all records are pooled in regional surveillance databases
- Find a way to integrate data from diverse practices
- Page 1.14
To pool and integrate data, need:
1) Encryption of data
2) HIPAA compliant and policies
3) Standard for data transmission
4) Standards for data definitions
5) Quality control and error checking
6) Regional and national surveillance databases
- HL7: The uniform envelope for digital communication
- Page 1.15
- Recommended steps for health promotion and disease prevention
- Detection of syndromes or problems
- Clinical guidelines
- Opportunities for distributed clinical research
- Page 1.17
- Creation of a National Health Information Infrastructure (NHI), which links all practices
and practitioners in the country
- Addresses error prevention
- Reduction in practice variation
- Reduced administration costs
- Enhanced efficiency
- Page 1.18
Cyclical Information Flow:
1) MD caring for patients
2) Uses EHR
3) Forward to national registries
4) Develop standard for prevention and treatment
5) Guidance from biomedical research
6) Researcher draws information from EHR or from pooled data
7) Translated into protocols, guidelines
8) New knowledge delivered back to clinicians
9) Integrate into EHR and order entry systems
- Page 1.19
- Telemedicine
- Use of email
- Web based disease management
- Page 1.19
Requirements for achieving the vision
1) Internet with high bandwidth and reliability
2) Decreased latency
3) Federal large scale networking
4) Education and training: Knowledge of the role computer can play in Health Care system
5) Organizational and Management change – Process Re-engineering
- Page 1.21
- Health institution to provide the standards, infrastructure, and resources
- A national initiative of cooperative planning and implementation is required
The use of Computers in Biomedicine
1) The application of computers in biomedicine
2) The concept of medical information
3) The structural features of medicine
- Page 1.22
- Medical computer science: Applying CS method to medical topics
- Information Science: Management of paper based and electronically stored information
o Aka cognitive science
- Information Theory: Physics of communication
- Biomedical Computing: Computers for purpose in biology
- Medical Informatics
- Biomedical Information Science and Technology Initiative (BISTI)
- Biomedical Informatics: The scientific field that deals with biomedical information, data and
knowledge – Their storage, retrieval and optimal use for problem solving and decision making
- Page 1.32
- Best viewed as a Biomedical science
- Use the results of past experience to understand, structure and encode objective and subjective
biomedical findings and make them suitable for processing
- Biomedical computing is an experimental science:
- Characterized by
1) posing questions
2) designing experiments
3) Performing analyzes
4) Using the information gained to design new experiments
- Goal:
1) Basic research: Search for new knowledge
2) Applications research: Use the knowledge for practical ends
- Page 1.32
Work in bioinformatics
1) Clinical Informatics: Clinical (Patient) care that demands patient oriented informatics
applications
2) Public health informatics (Population): Similar methods are generalized for applications to
populations of patients
3) Imaging Informatics (Tissue)
4) Bioinformatics (Molecular)
- Page 1.36
- MGH Utility Multi-Programming System (MUMPS language): developed for use in medicinal
applications
- Page 1.39
- Aspects of biomedical information include an essence of uncertainty: We can never know all
about a physiological process
- Page 1.39
- Low level processes
- In biomedicine, there are other higher level processes carried out in more complex objects such
as organism
- Page 1.40
- Biomedical Informatics includes computer applications that range from processing of very lower
level descriptions to processing of extremely high level ones
- Page 1.41
- Several global forces that affect biomedical computing:
1) New developments in Computer hardware and software
2) A gradual increase in the number of professionals who are trained in both clinical medicine
and biomedical informatics
3) Ongoing changes in health care financing designed to control the rate of growth of medical
expenditures
Chapter 2: Biomedical Data: Their Acquisition, Storage, and Use - Page 2.46
- All Medical care activities involve gathering, analyzing, or using data
- Medical datum: Any single observation of a patient
- Medical data are multiple observations
A patient data can be defined as:
1) The patient in question
2) The parameter being observed
3) The value of the parameter in question
4) The time of the observation
- Page 2.48
- The circumstances under which the data is observed
- Aka modifiers
- Uncertainty in the values of data
- Collect additional data to confirm or reject the concern
- Page 2.49
Types of Medical data:
1) Narrative data: Some narrative data are loosely coded with short hand conventions
2) Complete phrases have become loose standards of communication among medical
personnel
3) Numerical measurements: precision becomes important
4) Visual images
- Page 2.53
- Who collects data:
1) Doctors
2) Nurses
3) Office Staff
4) Specialists
5) Technological Devices
- Page 2.54
- Use of medical data:
1) To support the proper care of patient
2) Analysis of population
- Page 2.55
Uses of Medical Data
- A) Create the basis for the Historical record:
1) Patient’s history
2) Symptoms reported
3) Physical signs during PE
4) How have signs and symptoms changed?
5) What tests performed?
6) What medications taken?
7) Reasons behind the management decisions
- To answer 3 questions:
1) What was the nature of the disease
2) What was the treatment decision
3) What was the outcome of the treatment
- Central function to provide coordinated care to a patient over time
- Page 2.56
- B) Support communication among providers
o Importance of the central role of the medical records
o Serves as communication mechanism among physicians and other medical personnel
- Page 2.57
- C) Anticipate Future Health problems
- D) Record Standard preventive measures
o For interventions to prevent common or serious disorders
- E) Identify deviations from expected trends
- F) Provide a legal record
- G) Support clinical research
o Randomized clinical trial (RCT)
- Page 2.60
Weaknesses of the traditional medical record system:
1) Pragmatic and logistical issues:
a. Can I find the data I need when I need them?
b. Can I find the medical record in which they are recorded?
c. Can I find the data in the record?
d. Can I read and interpret the data
e. Can I find what I need quickly
f. Can I update the data reliably
- Page 2.62
2) Redundancy and Inefficiency
a. MD developed a variety of techniques that provide redundant recording to match
alternate modes of access
b. Inefficiency from tension between opposing goals in the design of reporting forms
- Page 2.63
3) Influence Clinical research
- Hard to flip through records for structured statistical analysis
- Page 2.64
- Retrospective Chart review vs prospective studies to collect future data relevant to the question
- Double blind studies: Patients and researchers do not know which treatment is being
administered
- Page 2.65
4) The passive nature of Paper records
- No warnings
- EHR system:
o Monitor their contents and generate warnings or advices for providers
o Provide automated quality control
o Provide feedback on deviation from standards
- Page 2.65
- Structure of Medical Data
- Need standardized Nomenclature:
- Use of coding systems:
1) ICD: International Classification of Disease: Used for discharge coding, and bills to insurance
2) Systemized Nomenclature of Pathology (SNOP)
3) Systemized Nomenclature of Medicine (SNOMED)
4) Current Procedural Terminology (CPT): By American Medical Association to produce bills for
services to patients
- Health Care professionals need standardized terms that can support pooling of data for analysis
and can provide criteria for determining changes for individual patients
- None will be completely satisfactory
- Some want more specific
- Some want more aggregation
- Page 2.69:
- Develop a unified medical language system (UMLS): ties together various vocabulary
- Central focus: Information base that constitutes the substance of medicine
- 3 terms:
1) Data: a single observation point that characterizes a relationship
- The value of a specific parameter for a particular object of a given point in time
2) Knowledge: derived through the formal or informal analysis of data
3) Information: encompass both organized data and knowledge
- Data are not information until they have been organized in some way for analysis or display
- Heuristics: A personal piece of knowledge that guides physicians in their decision making
- A database: A collection of individual observations without any summarizing analysis
- A knowledge base is a collection of facts, heuristics, and models that can be used for problem
solving
- Many decision support systems have been called knowledge base systems
- Page 2.70
Strategies of Medical data selection and use
1) Learn how to ask only the questions that are necessary to perform only the examination
components that are required
2) And to record only those data that will be pertinent in justifying the ongoing diagnostic
approach and in guiding the future management of the patient
3) Selectivity in data collection and recording
- Page 2.70
The Hypothetico deductive approach
1) Sequential staged data collection
2) Data collection
3) Hypothesis directed selection of the next appropriate data to be collected
4) As data are collected, they are added to the growing database of observations
5) Use to reformulate / refine the active hypotheses
6) Until one hypothesis reached a threshold of certainty
7) A management decision can be made
- Page 2.70
- Differential Diagnosis: Set of active hypotheses
- The selection process is heuristic
- Safety measures to avoid missing important issues:
1) Past medical history
2) Family history
3) Social history
4) Review of systems
- Focus hypothesis directed examination is augmented with general tests
- After asking questions, serves as the basis for a focused PE
- Treatment management plan can be developed
- Treatment itself is a datum point
- Page 2.73
- Chronic disease management: a cycle of treatment and observation for a long time
- Need to balance cost and risks
- Page 2.74
- Sensitivity: The likelihood that a given symptom can be observed in a patient with a given
disease
- Pathognomonic tests: they evoke a specific diagnosis but they also immediately prove it to be
true
- Specificity: An observation is not seen in patients who do not have the disease
- Prevalence of a disease: A measure of the frequency with which the disease can be found in a
population
- Page 2.75
- Predictive Value (PV): The post test probability that a disease is present based in the results of a
test
- PV+ = (sensitivity) x (prevalence) / ((sensitivity) x (prevalence) + (1 – specificity) x (1 –
prevalence))
- Post test probability is low if the prevalence of that disease is low
- Test is pathognomonic: When specificity is 100%
- PV+: One of many forms of Bayes Theorem: A rule for combining probabilistic data attributed to
Thomas Bayes
- Page 2.75
- Method for selecting questions and comparing test
- Page 2.76
- Computer and collection of data:
o By MD
o By Paid transcriptionist
o By Medical staff
o By device itself
o By patients
Chapter 19: Information Retrieval and Digital Libraries
- Page 19.660
- Information Retrieval (IR) is the field concerned with the acquisition, organization, and searching
of knowledge based Information
- Evolution:
- 1879: Dr John Show Billings created Index Medicus to help find relevant journal articles
- 1966: national Library of Medicine (NLM) unveiled an electronic version: The Medical Literature
Analysis and Retrieval System (MEDLARS)
- Page 19.661
- MEDLARS and MEDLINE stored only limited information for each article
- NLM assign to each article a number of terms from its Medical Subject Heading (MeSH)
- 1980: Full text databases emerge
- 1990: WWW: Medical information from multiple sources with various media extensions
available over the Internet
- Late 1990: NLM made all of its databases available to the world for free: The notion of digital
databases developed
- 21th century: Use of IR systems and digital library become mainstream
Biomedical Information:
1) Patient specific information applies to individual patients: about the health and disease of a
patient
- The patient’s medical record
- Page 19.662
2) Knowledge based Information: Derived from observational research:
a. Primary knowledge based Information
Original research that appears in journal
Reports the initial discovery of health knowledge with original data
b. Secondary knowledge based information
Writing that reviews the primary literature
Includes opinion based writings, clinical practices guidelines, systematic reviews,
and health information on web pages
Most common type of literature used by MD
Includes patient / consumer oriented health information on the web
- Libraries
1) Acquisition and maintenance of collections
2) Cataloging and classification of items
3) Individuals can go to seek information with assistance
4) Providing work or study place
- Page 19.663
4 states of information need
1) Unrecognized need: Clinician unaware of information need
2) Recognized need: Clinician aware of need but may or may not pursue it
3) Pursued need: Information seeking occurs but may or may not be successful
4) Satisfied need: Information seeking successful
- The information needs are not being met and IR application may help
- MD do not recognize that their knowledge is incomplete
- MD pursue only a minority of unanswered questions
- Unmet information on 2 questions / 3 patients
- The most frequent sources for answer to question was colleagues, followed by paper based
textbooks
- To lower the barrier to knowledge based information is to link it more directly with the context
of the patient in the EHR
- Most scientific journals are published electronically
- Authors have great incentives to maximize the accessibility of their published work
- Page 19.664
- The technical challenges to electronic scholarly publication have been replaced by economic and
political ones
- Someone has to pay the production costs
- Public funds the research, but individual libraries must buy it back from the publisher with the
copyright
1) Harneal proposed that authors pay the cost of production of manuscripts upfront. After the
paper is published, the manuscript become freely available
- E.g. Biomed Central (BMC): www.biomedcentral.com
2) PubMed Central (PMC, pubmedcentral.gov)
- Provides free access to published literature that allows publishers to maintain copyright and
keep paper on own server
- Lag time of 6 months
- Page 19.665
- Concerns:
1) Quality of information
a. Presence of inaccurate or out of date information
2) Readability: No health related site had readability below 10th grade level
3) Context Deficit: Fewer clear markers of the type of the document
- Readers of a specific page not aware of the context of a website---
- Page 19.666
- Ideal situation: partnership with patients and MD: MD primary source of recommendation for
online information
Criteria for a webpage
1) JAMA, Silberg, 1997:
- Contain the name affiliation, and credentials of the author
- References to the claims made
- Explicit listing of any conflict of interest
- Date of the most recent update
2) Health on the Net (HON) codes: a set of voluntary codes of conduct for health related
websites that adhere to the HON codes can display the HON logo
3) The American Accreditation HealthCare Commission (URAC): a process for accreditation
- 55 standards to support accreditations
- Page 19.666
- Evidence Based Medicine (EBM): a set of tools to inform clinical decision making
- Allows clinical experiences to be integrated with best clinical sciences
Process of EBM:
1) Phrasing a clinical question that is pertinent and answerable
2) Identifying evidence that address the question
3) Critically appraising the evidence to determine whether it applies to the patient
- Page 19.667
- 2 Types of Clinical questions:
1) Background questions ask for general knowledge about a disorder
2) Foreground questions ask for knowledge about managing a patient with a disorder
4 major foreground questions categories:
1) Therapy (or intervention): benefit of treatment or prevention
2) Diagnosis: Test Diagnosing disease
3) Harm: Detrimental health effects of a disease, environmental exposure, or medical
intervention
4) Prognosis: outcome of disease course
- First generation EBM: Focus on finding original studies in the primary literature and applying
critical appraisal
- Page 19.667
- Next generation EBM: Focus on the use of “synthesized” resources: The literature searching,
critical appraisal, and extraction of statistics operations are performed ahead of time
- More usable information resources such as advocated in systematic reviews
- Shaughnessy et al: The InfoMastery concept
- Church and Barnetts: Just in time information model
The 4S model of hierarchy of EBM resources:
1) The original Studies
2) Syntheses of those studies in systematic review
3) Synopses of these Syntheses
4) Systems that incorporate the knowledge of these studies for clinical decision support
- Page 19.668
Classify content categories:
1) Bibliographic content
o Includes the literature reference databases
o Aka bibliographic databases
o Consists of citations or pointers to the medical literature
o E.g. MEDLINE
o The current MEDLINE record contains up to 49 fields
o Available via PubMed
o Produced by the National Center for Biotechnology Information (NCBI)
o Ovid technologies and Aries System: License the content and provide value added
services
- Page 19.668
- Other more specialized databases
o Cumulative Index to Nursing and Allied Health Literature (CINAHL)
Covers nursing and allied health literature
o EMBASE: The electronic version of Excerpt Medica
European MEDLINE
More international focus
- Page 19.669
- Web Catalog: Web pages containing links to other web pages and sites
o HealthWeb.org
o HealthFindr.gov
o OMNI.ac.ak
o www.HON.ch/HONSelect
- Specialized registry: Literature reference databases, its indexes more diverse content than
scientific literature
- E.g. National guidelines clearing house (NGC)
o Produced by Agency for healthcare research and quality (AHRQ)
o Contains information about clinical practice guidelines
- Page 19.669
2) Full context content
- Consists of the online versions of books and periodicals
- Page 19.670
- BMI: British Medical Journal initiated an electronic long, paper short (ELPS) system that provides
on the Website supplemental materials that did not appear in the print version of the journal
- Allows direct linkage directly from the bibliographic databases to full text
- Most common secondary literature source is traditional textbooks
- Stat!Ref: Bundling of textbooks
- Harrison Online
- NCBI Bookshelf
- Page 19.671
3) Databases / Collection:
- Information housed in database management system
a. Image databases: collection of image from radiology, pathology
E.g. Visible human project genomic database
b. Genomics databases: information about gene sequencing, protein characteristics
NCBI
c. Citation databases: Scientific literature
Science citation index
Research Index uses autonomous citation indexing
d. EBM databases: collection of clinical evidence
Cochrane database of systematic review
Clinical Evidence
Up-To-Date
InfoPOEMS
PIER
e. Other databases
- Page 19.672:
4) Aggregation
- Have a wide variety of different types of information serving the diverse needs of users
- E.g. MEDLINE PLUS
- Merck Medicine
o Harrison’s Online
o MDConsult
o DXPlain
- Medweavers
o MEDLINE
o DXPlain
o WebCatalog
- Model Organism databases
- Page 19.674
Indexing 1) Manual indexing
2) Automated Indexing
- Page 19.675
- Concept: an idea that occurs in the world
- Term: string of one or more words that represent a concept
- Canonical Form: Preferred string of terms
- Synonyms: one or more terms that can represent a concept
- Controlled terminology: A list of terms that are canonical representation of the concepts
- Thesauri: contains relationships between terms
3 Categories:
1) Hierarchical: terms that are broader or narrower
a. Provides an overview of the structure of a thesaurus
b. Enhance searching
2) Synonymous: Terms that are synonyms
3) Related: Terms that are not synonymous or hierarchical but are somewhat related
- MeSH Terminology
1) 15 Trees
2) A large number of entry terms
3) Related terms
- Page 19.676
- Features of MeSH that makes documents more searchable
1) Subheadings: qualifiers of subject headings that narrow the focus of a term
2) Check Tags: MeSH terms that represents certain facets of medical studies: e.g. eye, gender
3) Publication type
- Other Indexes
o CINAHL subject headings
o EMTREE in EMBASE
- Page 19.670
- Need for linkage across different terminologies
- Unified Medical Language System (UMLS) Project
3 components of UMLS knowledge source
1) Metathesaurus
o All same terms linked as a concept
o Each term has one or more string
o One designated as the preferred form
o Preferred String of the preferred form is the canonical form of the concept
2) UMLS Semantic Network
3) Specialist Lexicon
- Page 19.678
- Manual Indexing 2 approaches:
1) Dublin Core Metadata Initiative
a. DCMI: www.dublincore.org
b. Applying Metadata to webpages and sites
2) Build directories of content
a. www.yahoo.com
- DCMI: First framework for Metadata on the web
- Page 19.679
- RDF: Resource Description Framework expressed in Extensible Markup Language (XML)
o Ability to express complex data
o Readability
o Tools to parse and extract data
o The preferred interchange format in the Clinical Document Architecture of the Health
Level 7 standard
- Manual Indexing: Inconsistence
- Page 19.680
- Automated Indexing: Word Indexing
- Take all consecutive alphanumeric sequences between white space with assign weighs
- Employ process to remove common words or conflate word to common forms
- Remove stop words
- Page 19.681
- Stop Word list: Negative dictionary
- Conflation of words to common form is done via stemming
o Ensure words with plurals are always indexed by stem form
- Term weighting: TF x IDF weighting
- Combines the inverse document frequency (IDF) and term frequency (TF)
- IDF: Inverse document frequency: The logarithm of the ratio of the total number of documents
to the number of documents in which the term occurs
- IDF = log (# of documents in DB / # of documents with term) + 1
- TF: Term frequency: Measure of the frequency with which a term occurs in a given document
- TF(term, document) = frequency of term in document
- Weight (term, document) = TF (term, document) x IDF (term)
- Page 19.681
- Use of link-based method
- Gives weight to page based on how often they were cited by other pages
- The Page Rank Algorithm (PR)
Word Indexing Limitations:
1) Synonymy: Different words may have the same meaning
o E.g. hypertension vs high blood pressure
2) Polysemy: same words may have different meanings or sense
3) Content: Words in a document may not reflect its focus
a. E.g. hypertension to refer to CHF
4) Context: Words take on meaning based on other words around them
5) Morphology: Words can have suffices that do not change the underlying meaning
6) Granularity: Queries may describe concepts at different level of hierarchy
- Page 19.682
Retrieval - 2 approaches:
1) Exact match searching: allows user precise control over the items retrieved
o Aka Boolean searching
o Aka set-based searching
o Exact match searching used with bibliographic databases
o Partial match approach used with full text searching
o First step:
Select terms to build sets
Other attributes may be selected
Combined with Boolean operators
Some allow using wild card character: aka truncation
2) Partial match searching: returns the user content ranked on how close it comes to the user’s
query
- Page 19.683
o Salton
o Most common use is with a query of a smaller number of words
o Aka natural language query
o Aka vector space model
o Aka relevance ranking: document ranked by their closeness of fit to query
o Aka lexical statistical retrieval
o Give each a score based on the sum of the weights of terms common to the document
and query
o Terms in document derive weight from TF x IDF calculation
o Document weight = sum of weight of term n query for all query terms X weight of term
in document
- Page 19.683
Retrieval Systems 1) Pubmed searches MEDLINE
a. Process input to identify MeSH terms, author names, common phrases, and journal
names
b. Allow the user to enter clinical queries: Subject terms are limited by search
statements designed to retrieve the best evidence based on principles of EBM
c. Approaches:
i. WebMEDLINE
ii. A great number use th Highwire system for online access
iii. ClinicalTrials.gov
- Page 19.687
Evaluation of IR system 1) Was the system used?
2) For what was the system used?
3) Were the users satisfied?
4) How well did they use the system
5) What factors are associated with successful or unsuccessful use of the system?
6) Did the system have an impact?
- Page 19.688
- Simpler Ways
1) System Oriented Evaluation: The focus of the evaluation is on the IR system
o Relevance based measures of recall and precision
o Rel: # of relevant documents
o Ret: # of retrieved documents
o Retrel: # of retrieved documents that are also relevant
o Recall = Retrel / Rel = what fraction of all the relevant documents have been obtained
from the database
o Most use Relative Recall: Denominator is the number of relevant documents identified
by multiple searches on the query
o Page 19.688
o Precision = Retrel / Ret = what fraction of retrieved documents are relevant
- Non ranking systems tend to retrieve a fixed set of documents
- Have fixed points of recall and precision
- Therefore, many evaluations of relevance ranking system will create a recall precision table that
identifies precision at various level of recall
- Page 19.689
- Text Retrieval Conference (TREC) organized by the US National Institute for Standards and
Technology
- For evaluation and presentation of results
- Alternatives: Measure using a more situational view of relevance
2) User Oriented Evaluation: Focus on the user
- Page 19.690
- Experienced clinicians and librarians archived comparable recall
- Librarian had statistically significant precision
- Novice clinicians had lower recall and precision
- The quality of documents is less important than whether the question is answered
- Page 19.692
- Research direction
- NLM: Biggest internal project is the Indexing Initiative
- Investigating new approaches to automated and semi-automated indexing
- Improved approaches to term weighing and language modeling
- Passage Retrieval: more weights based on local concentrations of query terms within them
- Query expansion: New terms from highly ranking documents are added to the query
automatically
- Page 19.693
- Digital Libraries’ full potential not realized
- Libraries: Maintain collection of published literature
- Quality control has been taken for granted
- Do not own artifact of the paper journal
- Have every URL linked to a uniform resource name (URN) that would be persistent
- The combination of URL and URN to provide persistent access to digital objects
- Page 19.684
- The use of Digital object Identifier (DOI): consisting of a prefix assigned by IDF to the publishing
entity and a suffix assigned and maintained by the entity
- Interoperability: How can resources with heterogeneous metadata be accessed?
- 3 levels of arguments:
1) Technical agreements over formats, protocols, and security procedures
2) Content agreement over the data and the semantic interpretation of its meta data
3) Organizational agreements over ground rules for access, preservation, payment,
authentication, etc
- Open Archives Initiative
- Page 19.695
- Preservation: Lots of Copies Keep Stuff Safe (LOCKSS) project
Chapter 6: System Design and Engineering in Health Care - Page 6.233
- System Success depends on the selection of adequate hardware and sufficient data storage and
data transmission capability
- The Software which defines how data are obtained, organized, and processed to yield
information
- Design systems that not only meet users requirements for information but also fit smoothly into
user’s everyday routines
Users
1) Health care professional
2) Administrators
3) Clerical Personnel
4) Operational Personnel
5) Professional Designers, Implementer, and Integrators
- Communication between HC and computing professionals in defining problems and developing
solutions that can be implemented within an industry
- Page 6.234
- Problem: practice in HC focuses on individual instances
- Flexibility can lead to an explosion of rules
- A system is an organized set of procedures for accomplishing a task
- Described in terms of
1) The problem to be solved
2) Data and knowledge required to address the problem
3) The process for transferring the internal input into the desired output
- A computer system contains both manual and automated process
- 3 components:
1) Hardware
2) Software
3) Customers
- Page 6.235
- Role of a computer is the conversion of data into information
- Organizational factors are crucial determinant of the success of a computer system within the
institution
- Page 6.236
Functions of a Computer System 1) Data acquisition and presentation
- Start with For automatic analysis of blood
- Patient monitoring system
- Medical Imaging Application
- Supply data directly to EMR
2) Record keeping and access
3) Communication and integration of information
- Essential for effective health care delivery
- Computers help by storing, transmitting, sharing, and displaying the data
- Page 6.238
- Patient record is the primary vehicle for communication of clinical information
- Paper based patient record:
a. Concentration of information in a single location
b. Prohibits simultaneous entry and access by multiple personnel
4) Surveillance
- Calling attention to significant events or situation
5) Information storage and retrieval
- Page 6.239
- Bibliographic retrieval Systems are essential of HIS
6) Data analysis and Presentation
7) Decision support
- All functions to support decision making
- Use population statistics or encode expert knowledge to assist MD in diagnosis and
treatment planning
- Assist in allocating nursing resources
- Use algorithmic, statistical, or AI techniques to provide advice about patient care
- § 4 key functions - · a. Administrative: - - Supporting clinical coding and documentation, - - authorization of procedures, and - - referrals - · b. Managing clinical complexity and details: - - Keeping patients on research and chemotherapy protocols; - - tracking orders, - - referrals follow-up, and - - preventive care - · c. Cost Control: - - Monitoring medication orders; - - avoiding duplicate or unnecessary tests - - d. Decision Support: - - Supporting clinical diagnosis and treatment plan processes; and
- - promoting the use of best practices, - - condition specific guidelines, and - - population based management
8) Education
- Page 6.240
- CDS that can explain their recommendations
- Page 6.240
- Identify the need for a computer System
o Improve the quality of care
o Lower the cost of care
o Improve access to care
o Collect information needed to document and evaluate the HC delivery process
- An information system cannot aid in decision making if the critical information is not available or
if the HC professionals do not know how to apply the information once they have it
- Computer cannot transform a poorly organized process into one that operates smoothly
- Page 6.241
- A careful workflow analysis before attempting computer based improvements
- Need -> techniques
- Techniques -> Apply need
1) Identify need
2) Identify the functions that fulfill the need
- Must start with problem definition
- Page 6.242
Need cost-benefit and cost effectiveness analysis
1) Define the problem
2) Establish Priorities of goals
3) Acquire a new system
4) Establish the system within the organization
- Page 6.243
- Requirement Analysis
- Sensitive to the various needs and probable concerns of the system’s intended uses
- Page 6.245
Why Fail? 1) Fail to account the key aspects of the way in which health professionals practice medicine
2) Cannot enter information along beside
3) MD had to enter the orders
4) Inflexibility of the system
5) Persistence negative biased caused by the earlier version
- Page 6.246
How to Design For Success 1) Analyze where the system is to fit into the existing workflow
2) Deciding what to purchase from a commercial system vendor and what to develop internally
3) Designing for actual customers
4) Involving those customers throughout the development
5) Planning for subsequent Changes
- Computer Systems will never be as flexible as human
- Page 6.247
- Circumscribe what a new system in a HC setting should encompass
- A successful system must take into account both the needs of the intended users and the
constraints under which these users function
- Page 6.248
- Commercial off the shelve software
- Major concern to ensure smooth transaction among software packages obtained from different
vendors
- Turnkey system: A vendor supplied system that requires only installation
- Custom Designed System
- Trade offs between compatibility with the conventions in the institution vs expense, delay, and
ongoing maintenance
- Compromise: Custom tailor a vendor system to the needs that are particular to one’s institution
- Question:
1) What is the extent of the support and maintenance
2) To what extent can the system be parameterized to the institution
- Page 6.249
- Technology Transfer
- Need to validate that the new system works under all conditions encountered in a clinical
practice
- System can recover from mechanical failure
- Work to link to other system
- Other systems may need to be adapted
- Difficulty of SW technology transfer
- Need to understand the data flow
A data flow diagram (DFD) 1) The source of data
2) The process for transforming the data
3) The points in the system where long term or short term data storage is required
4) Destination where reports are generated
5) Performance of quality assurance activities
- Page 6.251
- Need to obtain record information obtained informally
- All needed outputs to trace back to all required inputs
Data Storage - Ensure that the databases that come with commercial SW are open
- Documented, setup, and maintained
- Foreign applications can extract and contribute data to them
- Vendors schema must be adaptable
- Page 6.252
Water Fall model for development process 1) Requirements Analysis
2) Specification
3) Design
4) Implementation
5) Testing
6) Maintenance
- Page 6.253
- Water Fall model: Delay in such phased development process make this method inappropriate
- Page 6.234:
- A single module should not require more than a few implementers and not take more than a
few months to write an test
- HIS: A Hierarchy of nested and interrelated subsystems
- Page 6.235
- When all the components are ready, system integration takes place
- Riskiest Phase
- Testing takes as long as implementation
- Page 6.255
Alternative Methodologies
- Spiral Model: The group generates a simple prototype system by performing the 4 initial phases
rapidly (Boehm, 1998)
- Results presented to customers
- Assess it and expand and modify the requirements
- A second cycle starts
- After a few cycle, the prototype is made operational
- Any significant changes have to wait for another iteration
- A cycle = 3 to 6 months
- Little time spent on documentation
- Risks
1) Minimal Initial Prototype
o Not sufficient interest to the user
o No meaningful feedback
2) Does not attain sufficient functionality after several cycles
o Cannot be placed into services
3) An acceptable small system cannot be scaled the next level
- Page 6.256
- Water Sluice method: Easy processes are done first
- Business objects
- Object oriented programming focuses on the specification of small, reusable data structure and
associated fields called objects
- A suite of suitable objects: rapid composition of software modules
- Page 6.258
- The ability to interact with Remote system is essential
- External Services require different business arrangements depends on
1) Informational Services: Informal, broad public access
2) Business services: Contractual, user access
- For business services, the XML format provides a common syntactic technical infrastructure and
allows sharing of software tools
- System success depends on whether it meets the user’s informational needs and also how it
interacts with those users
- Must provide expert level advice
- Must be integrated into the daily routine of MD and other users
- Page 6.259
Parameters to consider For Successful System Implementation:
1) Quality and style of interface
2) Convenience
o Placement and number of PC
o Rapid logins
3) Speed and Response
4) Reliability: Redundant hardware and frequent data backup can minimize the loss of time
and data
5) Security
6) Integration
- Page 6.260
- Planning for Change
- Prototype Facilitate communication between computing personnel and users
- Formal Training courses to dispel the mystery of a new system
Chapter 5: Essential concept for Biomedical Computing
- Page 5.187
- Servers: Computers that share their resources with other computers and support the activities
of many users simultaneously within an enterprise
- Major servers are large mainframe computers
- Personal Computers
- Workstations: Interacting effectively with servers by integrating information from diverse
sources
- Page 5.188
- Terminals: Computers only for accessing servers and workstations
Computer architecture, e.g. a simple Von Newman machine
1) Central processing units (CPU)
2) Computer memories: store programs and data used actively by CPU
3) Storage devices: long term storage
4) Input and output devices
5) Communication equipments
6) Data buses: electrical pathways
- Page 5.189
- Hierarchical Organization: Primitive unit are combined to form base units
- Based units are assembled into registers
- Registers are assembled into large functional units that make up the CPU
- Binary bit: the logical atomic element of computers
- A byte = A sequence of 8 bits
- Foreign language users Unicode
- ASCII is a subset of Unicode
- Page 5.190
- A computer is an instruction follower: it fetches an instruction from memory and then executes
the instruction which usually is an instruction that requires the retrieval, manipulation, and
storage of data into memory or registers
- Page 5.191
Memory
1) ROM: Read Only Memory: Fixed memory
o Can be read, but not changed
o E.g BootStrap sequence: Set of instructions that is executed each time the computer is
started
2) RAM: random Access Memory
o Can be read and written into
- Word: A sequence of bits that can be accessed by the CPU as a unit
- Word Size
- For many medical applications: need to store more information that can be held in memory
- Long term storage
1) Active storage: Storage data that have long term validity and to retrieve with little delay
o E.g. medical record of a patient currently being treated
2) Archival storage: Store data for documentary
- Page 5.192
- Hard disks
- Floppy disks
- Memory sticks
- Flashcards
- Magnetic Tape
- CDs
- DVDs
- Page 5.194
- An alternative to physical transport is remote access to persistent storage by means of
communication networks
- Input Devices
o Keyboard
o Mouse
o Touch Screen
o Tactile Feedback: Computer controlled feedback
- Page 5.195
- Graphical interfaces and flowsheets are now common place in commercial clinical information
systems
- Image Libraries are used
- Text scanning devices
- Voice input
- Page 5.197
- Output Devices
o CRT deploy
o LCD Display
- Page 5.198
- A graphic screen is divided into a grid of picture elements called pixels
- The value of each pixel is associated with the level of intensity = grayscale
- For color display, the number of bits per pixel determined the contrast and color resolution of
an image
- Spatial resolution: The number of pixels per square inch
- Page 5.200
- Local data communication
- Bit Rates: Transmission speeds
- Overall bit rates of a communication link is a combination of the rate at which signals can be
transmitted and the efficiency with which digital information is encoded in the symbols
- Baud: one signal per second
- DSL
- ISDN
- Page 5.202
- Frame Relay: Network protocol designed for sending digital information over shared WAN
- Asynchronous Transfer Mode (ATM)
- A protocol designed for sending streams for small, fixed length cells of information over very
high speed dedicated connections
- LAN
- Page 5.203
- Broadband: Transmits multiple signals simultaneously
- Baseband: Transmits signals over a single set of wire
- Page 5.203
- Messages can be transmitted through the air by radio, microwave, IR, satellite signal, or line of
sight laser beam transmission
- Page 5.204
- Internet communication
- Transmission control protocol / internet protocol
- Computers hierarchical name management system called domain name system (DNS)
- Name servers convert a name into an IP address
- Page 5.206
3 conventions for composing internet names from segments
1) Functional convention
o Hierarchical segments from right to left, beginning with class identifiers
2) Geographical convention
o Beginning with top level country domain identifier
3) Attribute list address convention
o Names are composed of a sequence of attribute value pairs that specifies the
component needed to resolve the address
- Page 5.207
- Assembly language by assembler
- Combine sets of assembly instructions into macros
- Page 5.208
- Symbolic Programming Languages uses interpreter to convert and execute each statement
- Compiler translates all the statement at one time
- Type checking
- Page 5.209
- Sequence of statements are grouped into procedures
- These procedures are called with arguments
- Page 5.211
- Specialized languages can be used directly by non-programmers
- Page 5.211
Data Management - Applications must deal with large volume of varied data and manage them for persistence on
external storage
- Data can be viewed as:
1) Streams
2) Records
3) Hierarchy
- Page 5.212
- Operating System: A program that supervises, and controls the execution of all other programs
and that directs the operation of the hardware
- Applicator programs
- Page 5.213
- In multiusers system, users have simultaneous access to their jobs
- Multi programming: partition programs into pages, kept in temporary storage known as virtual
memory
- Page 5.215
Database Management System - Databases are collection of data, organized into fields, records, and files as well as descriptive
Meta Data
- Fields -> Records -> Files
- Meta data describes where in the record, specific data are stored and how the right data can be
located
- Data Independence: Keeping the application of one set of users independent from changes
made to application by another group
- Page 5.217
- The use of database technology combined with communication technology will enable HC
institutions to attain the benefits both of independent, specialized applications and of large
integrated databases
- Page 5.218
- Software for Network Communication
- Network stack: organize communications software within a machine
4 levels stack of TCP/IP
ISO Level TCP/IP Service Level
5 – 7 Applications: FTP, SMTP, TELNET
4 Transport: TCP, UDP
3 Network IP
1 – 2 Datalink and Physical Transport: Ethernet, Token Rings, Wireless
- Page 5.219
- Internet Protocols: Shared convention to standardize communications between machines
- Page 5.221
- XML: Extensible Markup Language
- Client Server Interaction: a generalization of interactions between a client machine and a server
machine
- Page 5.222
- Data Acquisition and signal processing
- Real time acquisition of data from the actual source by direct electronic connections to
instruments
- Signals are samples periodically and are converted to digital representation for storage and
processing
- Most naturally occurring signals are analog signals: Signals that vary continuously
- Digital Computers stores and process values in discrete values taken at discrete points and at
discrete times
- Analog to digital conversion (ADC)
o Sampling and recording
o The Continues value is observed (Sampled) at some instant and is rounded to the
nearest discrete unit
- Need one bit to distinguish between 2 levels
2 Parameters determined how closely the digital data represent the original analog signal
1) The precision: The degree to which a digital estimate of a signal matches the actual analog
value
2) Frequency: with which the signal is sampled
- Ranging and calibration of the instrument is necessary for accuracy
- The sampling rate: need to sample at least twice as frequently as the highest frequency
component that you need to observe as a signal
- Nyquist Frequency: the rate calculated by doubling the highest frequency
- Amount of noise in the signal
o The component of the acquired data that is not due to the specific phenomenon being
measured
- 3 techniques minimize the amount of noise in a signal
1) Shielding, isolation, and grounding of cables and instruments carrying analog signals
2) For robust transmission over long distance, analog signals can be converted into FM
o Conversion to digital forms provides the most robust transmissions
o Use of digital signal processing (DSP) chips
3) Filtering algorithms uses to reduce the effect of noise
- Page 5.226
Data and system Security - Privacy: The desire of a person to control disclosure of personal health and other information
- Confidentiality: Limit the further release of information
- Security: The protection of privacy and confidentiality through a collection of policies,
procedures, and safeguards
The security steps, serve 5 key functions:
1) Availability: Information available when needed
2) Accountability: Users responsible for their access to information
o Promoted by surveillance and by technical controls
o Through access audit trails
o Authentication and authorization
3) Perimeter Definition: allows the system to control the boundaries of trusted access to an
information system
o Requires that you know who your users are and how they are accessing the information
o Use cryptographic encoding to protect data
o Secret key cryptography: The same key is used to encrypt and decrypt the information
o Public Key cryptography: 2 keys are used, one to encrypt and a second to decrypt
o Certificates
4) Role limited access: enables access for personnel to only limited information
5) Comprehensibility and Control
Chapter 12: Electronic Health Record System - Page 12.447
Purpose of a patient record is to
1) Recall observations
2) Inform others
3) Instruct Students
4) Gain Knowledge
5) Monitor Performance
6) Justify Intervention
- To further the application of health science that improve the well-beings of patients, including
the conduct of research and public health activities that address population health
- Page 12.448
- An electronic health record is a repository of electronically maintained information about an
individual’s lifetime health status and health care
- Computer based patient-record system adds information management tools to help the users
organize, interpret, and react to data
Benefits of EHR
1) Flexible
2) Adaptable: Enter and display in various formats
3) Integrate Multimedia information
4) Accessibility
5) More Legible
6) Improve Completeness and Quality
7) Data can be reused
- Page 12.449
- Benefits depends on
1) Comprehensiveness of information
2) Duration of use and retention of data
3) Degree of structure of data
4) Ubiquity of access
- Disadvantages
1) Larger initial Investment
2) Human and organizational factors
3) MD takes time to learn the system and re-define work flow
4) Potential for failures and downtime
- Page 12.450
- History
- POMR: Problem Oriented Medical Record
- The Reimbursement model has shifted from a fee-for-service model
o Payers pay providers for all services deemed necessary towards a payment scheme
where provider are paid a fixed fee for a specific service
o Payer paid a fixed amount for services approved by the payer
- Page 12.452
- Ambulatory care records contain lengthy notes written by many different health are providers
Functional Components of EHR
1) Integrated view of Patient data
- Acquisition and organization of patients data are major challenges because of the complexity
and diversity of data and the large number and organizationally distributed sources of patient
data
- No unique national patient identifier exists in US for linking patient data obtained from many
sites
- Page 12.453
- Today most clinical data sources can deliver the clinical content as Health Level 7 (HL7)
messages
- Senders deviate from the standard
- Small amount of message tweaking
- Large amount of code mapping is required
- Interface engine providers a technical and translation buffer between systems manufactured by
different vendors
- Page 12.453
- Local terminologies represent major barriers to integration of EHR
- Code systems such as LIONC and SNOMED help overcome barriers
- Clinicians need integrated access to patient data
- MD need various view of data
- Page 12.455
2) Clinical Decision Support
- Most effective when provided at point of care, while the MD is formulating his assessment of
the patient’s condition and making ordering decisions
- Providing brief rationale with recommendation
- Solicitation of feedback when MD decides not to follow the recommendations
- Reminders and alerts can be raised during encounter
- Page 12.458
3) Clinical Order Entry
- System should present relevant information at the time of order entry
- Clinical alerts attached to a lab test result can also include suggestions for the appropriate
actions
- Warn about allergies and drug interactions when complete a mediation form
- Page 12.460
4) Provide access to knowledge resources
- At the time decisions or orders are being considered by MD
- E.g. PubMed
- OVID
- Up-to-Date
- Page 12.461
5) Integrated Communication and Reporting Support
- Messages electronically attached to EHR
- Help with patient handoffs
- Keep track of other patient encountered works
- Keep track of the time since the order was written
- Notify the MD when results are available
- Allow authorized provider to other information from another institute fro ER
- Facilitate efficient creation and transmission of reports
- Page 12.463
Issues for Computer based record system
1) Data Capture
a. Electronic Interfaces
Preferred method
Provide instant availability of data
b. Manual data entry
People must interpret or translate the data
Coding: data are classified and standardized
Takes time to translate human text into codes
Potential for coding errors
Errors difficult to detect
A variety of local coding system
Use Standard coding systems such as LIONC
Page 12.465
c. Physician entered data
MD’s notes can be entered via
i. Transcription of dictated or written notes
ii. Entry of data recorded on structured forms
iii. Direct data entry by MD
Dictation via voice recognition software
- Page 12.466
- Organizations are being encouraged to move to direct computerized physician order entry
(CPOE)
- EHR must apply validity checks
1) Range checks can detect or prevent entry of values that are out of range
2) Pattern checks
3) Computed checks
4) Consistency checks compare entered data
5) Delta checks between results
6) Spelling checks
- Page 12.466
Data Display
1) Flow sheets of patient data
- Organize patient data according to the time they were generated, emphasizing change over
time
2) Summaries and Abstracts
- Automated detection of adverse events
- Automated time-series events
3) Dynamic Displays
- Search tools to help MD locate data and specialized presentation formats to glean information
- Page 12.466
Query and Surveillance System
- Examine a patient’s EHR
- If the record meets pre-specified criteria, generate appropriate output
- Used for
1) Clinical Care: Preventive Care
- Recalled drugs notification
2) Clinical Research: Identify patients who met eligibility requirements for prospective clinical
trials
3) Retrospective Studies
- Contribute to research on a MD’s practice patients
- On the efficacy of tests and treatments
- On the toxicity of drugs
4) Administration
- Billing Data
- Claims data
- Page 12.471
Challenges Ahead for EHR
1) User’s Information Needs
2) User Interfaces: Must focus on MD’s unique information needs
3) Standards
a. Reduces development costs
b. Increase Integration
c. Facilitates the correction of meaningful aggregate data for quality improvements
HL7
DICOM: Radiology Images
LIONC for lab tests
SNOMED: By CHI: Consolidated Health Informatics
- Page 12.472
- Legal and Social Issues
- Need Federal law and policies
- HIPAA
- Costs and Benefits
- Difficult to determine
- Inability to measure accurately the actual costs and opportunity costs of using paper based
records
- Leadership
Chapter 7: Standards in Biomedical informatics - Page 7.265
- Standards: A set of rules that specify how to carry out a process or produce a product
- Page 7.266
- The need for Health Informatics standards
- Little coordination and sharing of patient data
- Standardize Identifier for individuals, health care providers, health plan, and employers
- Mechanism for issuing identifiers:
- CMS: National Provider Identifier (NPI)
o Payer ID for Health Care plans
o IRS Employer ID
- Person Identifier: Invasion of privacy
- Need Encoding of clinical knowledge using accepted standards
- Methods needed to transfer information from one to another
Standard Development Process
1) Ad Hoc Method
2) De Facto Method: Large vendor product make standard
3) Government Mandate method
4) Consensus Method: Group works in an open process to create a standard. E.g. HL7
- Page 7.269
- Process:
1) Identification stage: Need
2) Conceptualization Stage: Group formed to discuss
o What must the standard do?
o What is the scope?
o What will be its format?
3) Discussion stage
4) Writing for draft standard
o Page 7.271
o Open policy
o Open balloting policy
Negative ballots address
5) Early Implementation
o Acceptance and rate of implementation are important
o Need
a. Conformance
b. Certification
- Page 7.272
Standards Development Organizations
1) ANSI: American National Standards Institute
2) European Committee for Standardization Technical Committee 251 (CEN TC251)
o 4 working groups
a. Information Models
b. Terminology
c. Security, Safety and quality
d. Technology for interoperability
3) International Standards Organization Technical Committee 215
o Page 7.273
o Health Informatics
o 6 working groups
a. Health records and modeling coordination
b. Messaging and Communication
c. Health concept Representation
d. Security
e. Smart Cards
f. ePharmacy and Medicine
- Page 7.274
4) American Society for Testing and Materials (ASTM)
- Page 7.275
5) Health Care Informatics Standard Board
- Health Care Informatics Standard Planning Panel (HISPP)
- Identify Standards for:
a. Health care models and EHR
b. Interchange of health care data between organizations
c. Health care codes and terminology
d. Communication with diagnostic instruments
e. Communication of HC protocols
f. Privacy
g. Other Areas
- Page 7.276
6) Health Care Information and Management Systems Society
7) Computer Based Patient Record Institute (CPRI)
8) Integrating the Healthcare Enterprise (IHE)
o Stimulate the integration of HC Information resources
9) National Quality Form (NQF)
10) National Institute of Standards and Technology (NIST)
11) Workgroup for electronic data Interchange (WEDI)
a. Implementation of EDI
- Page 7.278
- Health Insurance Portability and Accountability Act of 1996 (HIPAA)
- Page 7.279
- Coded Terminologies, Nomenclature
- Purpose
1) Save system developers from reinventing the wheel
2) Facilitate exchange of data among systems
- To discuss coding systems:
1) Need to clarify the differences among a terminology, a vocabulary, and a nomenclature
2) Determine the basic use of the terminology: 2 Levels:
a. Abstraction: Examinations of the recorded data and then selection of items from a
terminology to label the data
b. Representation: process by which as much detail as possible is coded
- Page 7.280
- Must consider
1) Domain of Discourse: Must be good match with any standard selected for the purpose
2) Content of the Standard
3) The methods by which the terminology maintained method
- Specific Terminologies
1) International Classifications of Disease (ICD)
o Revised every 10 year
o ICD9 inadequate
o US published a set of clinical modifications (CM)
o ICD9-CM
- Page 7.283
2) DRG: Use in prospective payment in Medicare Program
- Provide a small number of codes for patients hospitalization
- DRG = Diagnosis related Groups
3) International Classification of Primary Care (ICPC)
- Post coordination of atomic terms
o Coding through the use of multiple codes as needed to describe the data
- Pre-coordination: Every type of pneumonia is assigned its own code
4) Current Protocol Terminology (CPT): a pre coordinated coding scheme for diagnostic and
therapeutic procedure
5) Diagnostic and statistical Manual of mental Disorder
6) Read Clinical Codes
7) SNOMED
- Clinical Terms and its predecessors
o SNOMED Clinical Terms (SNOMED CT)
8) Galen: Reference model for medical concepts using structures Meta Knowledge (SMK)
o Terms are defined through relationships to other terms
o Grammars to combine terms into sensible phrases
o Page 7.288
o To allow representation of information independent of the language being recorded and
of the data model used by EHR
9) Logical Observations, Identifiers, Names, and Codes (LOINC): For Tests and Observations
10) Nursing Terminologies
- North American Nursing Diagnosis Association (NANDA) codes
- Nursing Intervention Classification (NIC)
11) Drug Codes
- Drugs are classified according to the Anatomical Therapeutic Chemical (ATC)
- Page 7.289
- The National Drug Code by FDA
o Collaboration to produce a representational model for drug terms called RxNorm
- Page 7.291
12) Medical Subject headings (MeSH): Terminology by which the world medical literature is
indexed
13) Bioinformatics Terminologies
o Gene Ontology (GO)
o Develop standard ways of using these terminologies to encode data
o Distributed Annotation System
o Minimal Information about a Microarray experiment (MIME)
- Page 7.292
14) Unified Medical Language System (UMLS)
- Metasaurus: Contains over 100 different sources and relate synonyms across different sources
15) Interchange Registration of coding system
o Health Care Coding Scheme Designator (HCD)
- Page 7.292
Data Interchange Standards - By American Association for Medical Systems and Informatics (AAMSI)
- Developed standards for HL7 and IEEE Medical Data Interchange standard
- Page 7.296
- Purpose to permit one system, the sender, to transmit to another system, the receiver, all the
data required to accomplish a specific communication or transaction set in a precise,
unambiguous fashion
- Communication model: Open System Interconnection (OSI) reference model (ISO 7498-1) by ISO
- 7 Levels of requirements
1) Physical
2) Data Link
3) Network
4) Transport
5) Session
6) Presentation
o Deals with how data are formatted
o 2 philosophies
a. Position dependent format
Data content is specified and defined by position
b. Tagged field format
E.g SEX-M
7) Application
o Deals with semantics of data content specification of a transaction set
- Page 7.297
- A transaction set is defined for a particular event called a trigger event
Data Interchange Standards
1) Digital Imaging and Communications in Medicine (DICOM)
o For exchange of radiographic images
o Includes
Hardware Interface
Data Dictionary
A set of commands
o Data elements are organized within the data dictionary into related groups
o Groups and elements are numbered
o Each data element consists of its group element tag, its length, and its value
- Page 7.298
2) ASTM International
- E1238: Standard Specification for transferring Clinical observations between independent
systems
- Use position defined syntax
- E1467
- Page 7.299
3) HL7: Reflect the application (7th) Level of OSI Reference model
- Page 7.301
- Most widely implemented HC data messaging standard
- Message Based
- Use Event trigger model version 3.0
- Object oriented and Based on a reference Information Model (RIM)
- A collection of
a. Subject Area
b. Scenarios
c. Classes
d. Attributes
e. Use Cases
f. Actions
g. Trigger Events
h. Interactions
- To Provide a model for the creation of message specifications and messages for HL7
4) IEEE – MEDIX
- Standard of Medical Device Communications
- For medical Information Bus (MIB)
- For Bedside device in ICU, ER
5) National Council for Prescription Drug Program (NCPDP)
- For Pharmacy services
- 3 Standards:
a. Telecommunication Standard
b. SCRIPT Standard
c. Manufacturer Rebate Standard
6) ANSI X12
- For PO data
- Invoice data
- Claims, Benefits, Claims Payment
- Define Business Transactions in a formal Transaction sets
- Page 7.305
7) American Dental Association (ADA)
8) Uniform Code Council: Bar Codes
9) Health Industry Business Communication Council (HIBCC)
- Health Industry Bar Code (HIBC) standards
10) Electronic Data Interchange for Administration, Commerce, and Transport (EDIFACT)
- For Trade in goods and services
- Page 7.306
Directions For Standards
- Need standards for Queries
- Must support scripting and data entry mechanisms to ensure that a data system can properly
and accurately provide a response
- Require high speed query and response
- High computational speed
- Need To:
1) Negotiation among vendors interface systems
2) How closely the vendor’s implementation stick to the standard
3) Need certification
4) Lower standards change
- Page 7.309
- Future Directions:
1) Crossing the Quality Chasm
2) Consolidated Health Informatics (CHI) to establish a portfolio of existing clinical
terminologies and messaging standards to build interoperable federal health data systems
- Develop National health care Informatics Infrastructure to
1) Improve Patient safety and quality
2) Detect Bioterrorism
3) Enhance the efficiency of HC System
- Problem: Conflict between a standard and the opportunity for a vendor to use creativity in a
product to enhance sales
- Standards should encourage Creativity
o Use of Mouse
o Use of visual objects
o Use of Icons
- Standards need to include all types of data representation
- Specifying the location of data
- Rules for the creation of data
- Tighter coupling of data
Chapter 10: Ethics and Health Informatics: Users, Standards, and
Outcomes - Page 10.381
- There is less evidence that existing clinical expert systems can improve patient care in typical
practice settings at an acceptable cost in time and money
- Human Cognitive processes for diagnosis should not be trumped by computers
1) MD to use computer based tool responsibility
2) Not abrogate their clinical judgment reflexively
- For Error Avoidance and
- Capturing Evolving Standards
- Ethically optimized action
- Page 10.382
Appropriate Users and Educational Standards
- To use a diagnostic decision support system, MD must be able to recognize
1) When the computer program has erred
2) When it is accurate
3) What the Output means
4) How it should be interpreted
- Page 10.383
- MD must offer context, comfort, and hope
- Ethical Principles
1) A computer system should be used in clinical practice only after appropriate evaluation of
its efficiency and the documentations that it performs its intended task at an acceptable
cost and time and money
2) Users should be Health Professionals who are qualified to address the question at hand on
the basis of their licensure, Clinical training, and experience
- Software systems should be used to augment or supplement, rather than to replace or supplant
individuals decision makings
3) All uses of informatics tools should be preceded by adequate training and instruction, which
should include review of all available forms of previous product evaluations
- Page 10.384
- Professional Patient Relationship
- Trust
- Quality Standards should stimulate scientific progress and innovations while safeguarding
against system error and abuse
- Page 10.384
- Progressive Caution: Scientific and technical standards are able to stimulate progress while
taking a caution or conservative stance toward permissible risk in patient care
- Page 10.385
10 Criteria for systems scrutiny
1) Does the system work as designed
2) Is it used as anticipated
3) Does it produce the desired result
4) Does it work better than the procedures it replaced
5) Is it cost effective
6) How well have individuals been trained to use it
7) What are the anticipated long term effects on how department interact
8) What are the long term effects on the delivery of medical care
9) Will the system have an impact on control in the organization
10) To what extend do effects depend on practice setting?
- Page 10.386
- Balance between
1) Free access to information
2) Protection of patients’ privacy and confidentiality
- Privacy: Applies to People
- Confidentiality: Applies to Information
- Page 10.387
- No absolute confidentiality
- Patient Medical records give to HC providers for access
- Contacts with contagious disease to identify
- To determine naturally history of disease by pooling data
- The need for robust syndrome surveillance for adequate bioterrorism preparedness
- Page 10.388
- To restrict inappropriate access to electronic records
1) Technological methods: Computer security
2) Policy Approaches: Create security and confidentiality committees and establish education
and training programs
- Page 10.389
- Establish safeguards that optimize the research ethically
1) Establish mechanisms to anonymize the information in individual records
2) Use of institutional panels, such as medical record committees
- Page 10.390
- Social challenges and ethical obligations
- The increasing unwillingness of governments and insurers to pay for interventions and therapies
that do not work
- Page 10.391
- Prognostic Scoring Systems that use physiologic and mortality data to compare new critical care
patients with thousands of previous patients
- Allow hospitals to track their performance of their critical care units, by comparing outcomes
among years
- Identify ways to improve care of patients
- Page 10.392
- Cannot deny therapy for patients meeting criteria
- 2 uses
1) Prognostic Scoring system to justify termination of treatment to conserve resources
2) Diagnostic expert system to deny a MD reimbursement for procedures deemed
inappropriate
3 reasons to use clinical computer programs to guide policy
1) Human cognition is superior to machine intelligence
2) Decisions are value laden and made relative to treatment goals
- To improve quality of life
3) May include individuals in groups they resemble but do not belong
- Page 10.393
- User computers to improve professional
- Patient Relationship
- Page 10.394
- Consumer Health informatics have misinformation
- Internet resource:
1) Peer Review
2) Online Consultation
- Especially important in the context of telemedicine
3) Support Group
- Page 10.395
- Legal: Deal with practices of regulation of morality or behaviors
- Ethical: Determine what is good and which behaviors are correct in accordance with higher
principles
- Liability Under Tort Law
Clinical Software applications can be viewed as a product or a service
1) Negligence Theory: sue because of making decision using a product or making decision
without using a product
2) Product Liability:
- Harm caused by defective products
- 3 conditions:
1) Product purchased and used
2) Purchaser suffer physical harm as a result of defect
3) Product shown to be “unreasonably dangerous” demonstrate cause of injury
- Negligence theory allows for adverse outcome
- No exception for strict product liability
- Page 10.397
- HIPAA for Privacy and Confidentiality
- Page 10.398
- Facts in database may not be copyrightable
- Contents for materials on WWW usually in public domain
- Page 10.399
Recommendations
1) Recognition of 4 categories of clinical system risks and 4 classes of monitoring and
regulatory action that can be applied
2) Local oversight of clinical software systems through software oversight committees
3) Adoption by HC IS developers of a code of good business practices
4) Recognition that budgetary, logistics, and other constraints limit the type and number of
system that the FDA can regulate effectively
5) Concentration of FDA Resolution on those systems posing highest clinical risks, with limited
opportunities for competent human intervention, and FDA exemption of most other clinical
software systems
- Page 10.399
Principles for Decision making
1) Specially Trained human are best able to provide HC for other human
- Software should not overrule a human decision
2) MD who used Informatics Tools be clinically qualified and trained in using Software products
3) Tools evaluated and regulated
4) HC tools evaluated on their influences on institutions, institutional cultures, and workplace
social forces
5) Ethical obligations extend to non clinicians
6) Educational programs and security measures essential for privacy and confidentiality
Adequate Oversight maintain to optimize ethical use of EMR for research
Chapter 3: Biomedical Decision Making Probabilistic Clinical Reasoning
- Page 3.80
- Medical practice is decision making
- Page 3.82
- Many decisions are made on the basis of knowledge that has been gained through collective
experience
- MD must rely on empirical knowledge of associations between symptoms and disease to
evaluate a problem
- A decision that is based on these imperfect association will be uncertain
- Page 3.83
- Clinical data are imperfect
- All clinical data are uncertain
- The words “probable” and “highly likely” have different meaning to different people
- Page 3.84
- The odds are an alternate way to express a probability.
- The use of probability or odds as an expression of uncertainty avoids the ambiguities inherent in
common descriptive terms
Diagnostic Process Stages
1) Making an initial judgment about whether a patient is likely to have a disease
- Pretest probability: The estimated probability, made before further information is obtained
2) Gather more information by performing a diagnostic test
- The more a test reduces uncertainty, the more useful it is
- Page 3.85
3) Update the initial probability estimate
- The posterior P(D)
- Post Test P(D)
- To calculate post test P(D), we must know the pretest P(D), and the S&S of the test
- Page 3.86
- P[A] = Probability of event A
- P[A,B] = Probability of event A and B occurring together
- The P(D) of 2 independent events A and B both occurring
- P[A,B] = p[A] x p[B]
- P[A|B] = Conditional P(D) of event A given event B
- The p(D) that event A will occur given that event B is known to occur
- E.g. p[blood clot| swollen leg]
- The pretest p(D) is the prevalence of blood clots in the leg in a population
- Page 3.86
- Subjective p(D) assessment
- Most assessments are based on personal experience
- Page 3.87
- People rely on discrete, often unconscious mental processes
Cognitive heuristics
- A process by which we learn, recall, or process information
1) Representativeness – Probabilities are judged by the degree to which A is representative of B
- E.g. blood clot in swollen leg
- Difficulties with the use when
a. Disease is rare
b. MD previous experience is a typical
c. Patient’s clinical profile is a typical
d. The p(D) of certain findings depends on whether other findings are present
2) Availability: Events more easily remember are judged more probable
- E.g. MD cared for patient who had swollen leg and died from blood clot
3) Anchoring and Adjustment: Makes an initial p(D) estimate (the anchor) and then adjust the
estimate based on further information
- E.g. Initial P[Heart disease] = 0.5
- Adjust because patient’s family had heart disease
- Mistake to adjust the initial estimate insufficiently in light of new information
- Page 3.88
- Public research results can serve as a guide for more objective estimate of p(D)
- Prevalence: The frequency of an event in a population
- Place patients into a clinical subgroup in which the p(D) of disease is known
- Use clinical prediction rules to help MD assign patients to well-defined subgroups
o Developed from systematic study of patients who have a particular diagnostic problem
o Signs and symptoms are assigned numerical weights
o The total score places a patient in a subgroup with a known probability of disease
o Subject to error because of bias in the studies on which the estimates are based
- Page 3.90
- E.g. Prevalence of disease in a specialist population is much higher than in primary care practice
- Aka Referral bias
- MD may use subjective methods to adjust his estimate further based on other specific
information about the patient
- Page 3.91
- Determine criteria for deciding whether a result is normal or abnormal
- A test result greater than 2 standard deviations above the mean is reported as abnormal
- A test below the cutoff is reported as normal
- An ideal test would have no values at which the distribution of diseased and nondiseased people
overlap
- True Positive (TP): Positive results obtained for patients with disease
- True Negative (TN): Negative results for patients without the disease
- False Positive (FP): Positive results for patients without disease
- False Negative (FN): Negative results for patients with disease
- Page 3.92
- 2 x 2 Contingency table
Disease No Disease
Positive Result TP FP
Negative Results FN TN
- TP + FN = Total # of diseased
- FP + TN = Total # of non diseased
- TP + FP = Positive results
- FN + TN = Negative results
- A perfect test would have no FN or FP
- Page 3.93
Measure of Test performances
1) Measures of concordance: Measure of agreement between test
a. TP and TN
2) Measures of discordance: Measure of disagreement between tests
a. FP and FN
- TPR = Sensitivity: True Positive Rate
- = TPR
- TPR = The likelihood that a diseased patient has a positive test
- TPR = p[positive test | disease]
- TPR = # of diseased patients with positive / Total # of diseased patients
- TPR = TP / (TP + FN)
- TNR = Specificity = True Negative Rate
- = TNR
- TNR = The likelihood that a nondiseased patient has a negative test result
- TNR = p[negative test | no disease]
- TNR = # of non diseased patients with negative test / Total # of non diseased patients
- TNR = TN / (TN + FP)
- FPR = False Positive Rate
- FPR = The likelihood that a non diseased patient has a positive test
- FPR = FP / (TN + FP)
- FNR = False Negative Rate
- FNR = The likelihood that a diseased patient has a negative test
- FNR = FN / (TP + FN)
- Page 3.94
- TPR + FNR = 1
- TNR + FPR = 1
- Page 3.95
- The calculated value of S&S for a continuous valued test depend on the particular cutoff value
chosen to distinguish normal and abnormal results
- Higher cutoff point
o Decrease false positive
o Increase False Negative
o Test become More specific
o Test become less sensitive
- Lower cutoff point
o Decrease false negative
o Increase false positive
o Increase sensitivity
o Decrease specificity
- Need to decide if it is better to have more false negative (missed cases) or False positive
- Minimize false negative If the disease is serious and if life therapy is available
o Don’t want to miss anyone
- Minimize false positive if disease is not serious and therapy is dangerous
o Don’t want to apply unnecessary test
- S&S are criteria on when to call the test abnormal
- ROC curve = Receiver Operating characteristic curve
- Plot the test’s sensitivity against 1 – specificity
- TPR vs FPR
- Any point on an ROC curve corresponds to the test S&S for a given threshold of abnormality
- A test has better discriminating power than a competing test if its ROC curve lies above that of
the other test
- Test B is more discriminating than test A when its specificity is greater than test A specificity, for
any level of sensitivity
- MD choose tests depend on
o Cost
o Risk
o Discomfort
o Delay
- Page 3.96
- Choose test with the highest S&S when other factors are the same
- Gold Standard test: A procedure that is used to define unequivocally the presence or absence of
disease
- E.g Biopsy of tissue
- Index test: The test whose discrimination is being measured
- Gold Standard test is usually more expensive, riskier, or difficult to perform
Index test performed in 2 groups
1) Study population: Patients in whom test discrimination is measured and reported
2) Clinically relevant population: Patients in whom a test typically is used
- Page 3.97
- Referral Bias: Diseased patients are more likely to be included in studies than are non diseased
patients
- Problem arises when the TPR and TNR, measured in the study population, do not apply to the
clinically relevant population
Because of
1) Spectrum bias: when the study population includes only individuals who have advanced
disease and healthy volunteers
- The clinically relevant population will contain more cases of early disease that are more likely to
be missed by the index test (FN) when compare to the study population
- Therefore, the study population will have an artificially low FNR: Which produces an artificially
high sensitivity TPR (TPR = 1 – FNR)
- Health patients in study population will have lower FPR, therefore, the specificity (TNR) will be
overestimate (TNR = 1 – FPR)
2) Test Referral Bias: A positive index test is a criteria for ordering the gold standard test
- The study population has higher % of patients with disease
- Therefore, TPR increases -> FN decreases, TN decreases
- Page 3.98
3) Test interpretation bias: When the interpretation of the index test affects that of the gold
standard test or vice versa
- Person interpreting the index test should be unaware of the results of the gold standard test
- To counter these bias: Adjust TPR and TNR when they applied to a new population
- All biases result in a higher TPR that is higher in the study population
- Adjust TPR (Sensitivity) downward when apply to a new population
- Adjustment of TNR (specificity) downward for spectrum bias and test interpretation bias
- Adjustment of TNR upward for test referral bias
- Page 3.98
- Assess the quality of the studies and to use the estimate form the highest quality studies
- Perform a meta-analysis: A study that combine quantitatively the estimates from individual
studies to develop a summary ROC curve
- Page 3.99
- Bayes’ Theorem is a quantitative method for calculating post-test p(D) using the pretest p(D)
and the S&S of the test
- P[D|R] = p[D] x p[R|D] / (p[D] x p[R|D] + p[-D] x p[R|-D])
- P[D|+] = p[D] x TPR / (p[D] x TPR + (1 – p[D]) x FPR)
- P[D|-] = p[D] x FNR / (p[D] x FNR + (1 – p[D]) x TNR)
- Page 3.100
- Odds = p / (1 – p)
- P = odds / (1 + odds)
- P[D] = 0.75, P[-D] = 0.25
- Diseased = ¾
- No Disease = ¼
- Post test odds = pretest odds x Likelihood ratio
- Odd Ratio Form of Bayes Theorem: P[D|R] / P[-D|R] = P[D] / P[-D] x P[R|D] / P [R|-D]
- LR = P[R|D] / P[R|-D]
- LR = probability of result in diseased / probability of results in non diseased
- LR+ = probability of test positive in diseased people / probability of test positive in non diseased
people
- LR+ = TPR / FPR
- Page 3.101
- In a test that discriminates well between disease and non disease, the TPR will be high, the FPR
will be low
- LR+ will be much greater than 1
- A LR of 1 has no value
- LR - = P(D) that test is negative in diseased people / p(D) that test is negative in non-diseased
people
- LR- = FNR / TNR
- A desirable test will have low FNR and a high TNR
- LR- will be much less than 1
- Steps to
1) Find pretest odds = Pretest P(D) / (1 – Pretest P(D))
2) Find LR+ for test = TPR / FPR
3) Post test odds = Pretest odds x LR+
4) Post test P(D) = Post test odds / (1 + Post test odds)
- Page 3.102
- Positive predictive value is the likelihood that a patient has a positive test result also has disease
- PV+ = # of diseased patients with positive test / Total # of patients with positive test
- PV+ = TP / (TP + FP)
- Negative Predictive Value
- PV- = # of nondisease patients with negative test / total number of patients with negative test
- PV- = TN / (TN + FN)
- Page 3.103
- The PV of a test depends on the prevalence of disease in the study population
- Prevalence = (TP + FN) / Total # of patients
- PV cannot be generalized into a new population because the prevalence of the disease may
differ between the 2 populations
- Bayes theorem provides a method for calculation of disease for any prior probability
- Implication of Bayes Theorem for test interpretation
- The Post test Probability of disease increases as the pretest p(D) of disease increases
- The 45 degree line = a test in which the pretest and posttest p(D) are equal (LR=1) a test that is
useless
- For Positive test results:
o At high pretest p(D), the post test p(D) is slightly higher
o At low pretest p(D), the post test p(D) is much higher than pretest p(D)
- For negative test results:
o At high pretest p(D), the post test p(D) is much lower than the pretest p(D)
o Little effect at low pretest P(D)
- Page 3.105
- If Pretest p(D) is very low, a positive test result can raise post test p(D) into only the
intermediate range
- Exception: Test with very high specificity (0.99): A positive test is convincing
- If Pretest p(DP is very high, a negative test only lower the post test p(D) to mid range
- Test Specificity affects primarily the interpretation of a positive test
- Test Sensitivity affects primarily the interpretation of a negative test
- An increase in the specificity of a test, markedly changes the post test p(D) if the test is positive
but has little effect on the post test p(D) if the test is negative
*** To Rule in a Diagnosis, Choose a test with high Specificity or a high LR+ ***
*** To Rule Out a disease, choose a test with a high sensitivity or a high LR- ***
Cautions using Bayes Theorem
- Common problems
1) Inaccurate estimate of pretest p(D)
2) Faulty application of test performance measures
3) Violation of the assumptions of conditional independence and of mutual exclusivity
- Page 3.106
- Accuracy of the estimated pretest p(D) is increased by proper use of published prevalence rate,
heuristics, and clinical prediction rules
4) Apply published values for the test S&S, or LR, without paying attention to the possible
effects of bias in the studies
5) Use Bayes’ theorem to interpret a sequence of test
- Valid only if the 2 tests are conditionally independent
- When the p(D) of a result on 2nd test does not depend on the result of the 1st test
6) Assume that all test abnormalities result from one and only one disease process
- Page 3.106
- Shows the post test probabilities for tests with varying specificities (TNR)
- Changes in specificity produce large changes in the top family of the curves (positive test results),
but little changes in the bottom family of the curves (negative test results)
- Increase in specificity changes the post test probability if the test is positive
- Increase in specificity has little effect on post test probability if the test is negative
- Spin: Choose tests with High Specificity Rules In a Disease
- Shows the post test probabilities for tests with varying sensitivities
- Changes in sensitivity produces large changes in the bottom family of the curves (negative test
results) but little effect on the top family of curve (positive test results)
- Snout: Choose tests with high sensitivity to rule out a disease
- Page 3.107
Expected Value Decision Making
- Characterize each gamble by a number
- Use the number to compare the gambling
- Utility is the name given to a measure of preference that has a desirable property for decision
making: the gamble with the highest utility should be prefer
- Therapy A mean survival = 2.3 years
- Therapy B = 3.1 years
- Therefore choose therapy B
- Page 3.109
- Chance node : Events that are under the control of chance
- O---
- The outcome of a chance event can be represented by the expected value of the chance node
- Page 3.110
To use expected value decision making
1) Calculate the expected value of each decision alternative
2) Pick the alternative with the highest expected value
- Page 3.111
4 steps in decision analysis 1) Create a decision tree
2) Calculate the expected value of each decision alternative
3) Choose the decision alternative with the highest expected value
4) Use sensitivity analysis to test the conclusion of analysis
- Page 3.112
- Quality of Life Year (QALY)
- Duration of survival in good health
- “How many years with normal mobility is equivalent in value to 10 years in current state of
disability
- Page 3.114
- Calculate the p(D) of death
- Expected value at A = 0.05 x 0 + 0.05 x 3
- Average out the chance node
- 7.7 years > 6 year
- Therefore do surgery
- Results represent the outcomes that would occur on average in a population of patients who
have similar utilities
- Page 3.115
Represent patient’s preferences with utilities
- Utility of a health state is the quantitative measure of the desirability of a health state from a
patient’s perspective
- 0 = death
- 1 = ideal health
1) The standard gamble technique has the strongest theoretical bias of the various approaches
to utility assessment
- Choose between infection and a gamble with chance of ideal health or immediate death
2) Time trade off technique
- Determine the length of time in a better state of health equivalent to a long period of infection
- Risk Neutral
3) Visual Analog Scale: Rate quality of life with health outcome form 0 to 100
- Sensitivity analysis: A test of the validity of the conclusion of an analysis over a wide range of
assumptions about the p(D) and the values
- Do my conclusions regarding the preferred choice change when the p(D) and outcome estimates
are assigned values that lie within an reasonable range?
- Page 3.118
- Representation of long term outcomes with Markov Models
- Page 3.120
Decision whether to treat, test, or do nothing
1) Determine the treatment threshold p(D) of disease
2) Determine the pretest p(D) of disease
3) Decide whether a test result could affect your decision to treat
- Treatment Threshold p(D) of disease = p(D) of disease at which you should be indifferent
between treating and not treating
- Page 3.122
- P* = H / (H+B)
- P* = Treatment threshold p(D)
- H = Harm associated with treatment of patient without disease
- B = Benefit associated with treatment of diseased patient
- B = U [ D, treat] – U[D, not treat]
- H = U [-D, no treat] – U [-D, treat]
- Page 3.122
- Do not order a test unless it would change your management of the patient
- Order a test only if it cause the p(D) of disease to cross the treatment threshold
- Page 3.124
- Influence Diagrams
- Obtain PCR? Y/N -> PCR Result <- HIV Status -> QALE
- Page 3.128
- The method for evaluating is sensitivity analysis: examine any variable to see whether its value is
critical to the final recommended design
Decision models
1) Help to structure guideline development problems
2) Incorporate patient’s preferences
3) Tailor guidelines for specific clinical populations
4) Web based decision model provide distributed decision support for guideline developers an
users on the Web
Chapter 16: Patient Care System (1 slide) - Page 16.564
- Patient Care is an interdisciplinary process centered on the core recipient in the content of HC
disciplines according to patient needs
Complicating Factors for caregivers
1) Simultaneous attention to multiple aspects of the patient
2) Interdisciplinary care: Care provided by each discipline
- Collaboration requires exquisite communication and feedback
3) Coordination among multiple caregivers
- Sometimes by a case manager
4) Each caregiver must consider the competing standards of all the patients and exigencies of
all other HC professionals
- Must plan for optimal scheduled treatment
- When unexpected needs arises, must set priorities
5) Caregivers responsible for indirect care activities: Teaching, Meetings
- Page 16.568
Information to support patient care
1) Who is involved in the care of patient
2) What information does each professional require to make decision
3) For where, when, and in what form does the information came?
4) What information does each professional generate? Where, when and in what form is it
needed?
3 information categories:
1) Patient specific data: data about a particular patient from a variety of sources
2) Agency specify data: data relevant to the specific organization
3) Domain information and knowledge: specific to HC disciplines
4 types of Information Processes
1) Data Acquisition
2) Data Storage
3) Data Transformation
4) Data Presentation
- Need for Presentation at the point of patient care
- Integration of up to the minute patient specific data with agency specific guidelines
- Page 16.571
- Information systems must be geared to the needs of all the professional involved in care
- Systems aid the coordination of interdisciplinary services and planning of caregivers work
activities
- 1960: Technician Medical Information System (TMIS)
- Development and evolution of systems that support clinical decision making
- HELP Systems: Initially for MD during process of care
- Subsequently support nursing care decision
- Aggregate data for research
- Page 16.572
Today’s patient care system
1) Decision support
2) Integration of Information
3) Care planning and documentation
4) Organization of MD’s workflow
5) Support for Care management
Societal forces have influenced the decision and implementation of patient care system
1) Delivery System Structure: Shift from single institution structure to the integrated delivery
network
2) Professional Practice Models: Team Nursing + Nursing Care Plan
- Shift to interdisciplinary care approaches across the continuum of care
- Advanced Practice nurses taking on functions by MDs
- MD practice model shifted from single MD to complex consternation of provider organization
- For issues such as
a. Locations of medical records
b. Control of practice patterns of MD
c. Data Reporting requirement
- Page 16.574
- Creation of a single patient problem list among HC providers
3) Payer Model
- Shift from fee for service to prospective payment to capitation
- NC need to comply with HIPAA
4) Quality Focus
- Old: Prospective chart audit
- To continuous quality improvement techniques
- to Quality Management techniques
- to patient care system based approaches such as
a. Critical Path
b. Practice Guidelines
c. Alerts
d. Reminders
- For benchmark purposes
- Report processes to regulatory bodies
Patient Care Systems:
- PROMIS (Problem Oriented Medical Information System)
- TRIMUS
- HELP
- DHCP
- COSTAR
- TMR
- Page 16.575
- The most commonly used system are those that support nursing care planning and
documentation
1) Systems to support capture of MD orders
2) Communication with Pharmacy
3) Reporting of lab results
4) Merge MD orders with nursing plan
- Early ambulatory care
- Paper based Patient encounter forms
System Uses:
1) Entry of structured information
2) Retrieval of reports
3) Decision support
4) Alerts to Remind MD about needed care
5) Avoid contraindication
6) Avoid unnecessary lab analysis
7) Facilitate Communication
New systems need to
1) Aggregate data across patients
2) Query data about a subset of patients
3) Use data collected for clinical purposes
4) For administration
- Page 16.576
- Collect Once, Use Many Times
Data Aggregate to:
1) Analyze for Admin reports
2) Quality Improvements
3) Clinical Research
4) Required patient safety and Public Health reporting
- Clinical Pathways: Coordination of the information and service of the various clinical disciplines
into integrated records and plans
Best Patient care Systems:
1) Broadscale computer based results reporting
2) Order entry for medications
3) Documentation of MD notes
4) Closed Loop medication systems use technologies such as bar codes and decision support to
guard against errors
5) Offer “Best Practice” guidelines
6) Use Knowledge bases and patient data to access orders for potential contraindication
7) Offer point and click access to knowledge summaries and publications
- E.g.
- LDSH
- BWH
- WHH
- QMC
- VARS
- Page 16.578
- Challenges to transition from self contained system to Integrated Systems
Cornerstone of good patient care system:
1) Ability to capture data in the process of care
2) Store data
3) Aggregate data
4) Analyze
5) Produce reports for knowledge of quality, effectiveness, and costs for the bases of improved
clinical processes
- Key is data standards
- HC professionals must be informatics competent
- Page 16.579
4 Categories: Typology of Science in Medical Informatics
1) Formulating models for acquisition, representation, processing, display, or transmission of
biomedical information or knowledge
- By standards development organization (SDO)
2) Development of innovative system
- New systems take advantage of information entered in one content for use in other content
- E.g. Use information for order entry to prepare discharge orders
- E.g. Low cost bed side workstations
- Page 16.579
- Need to make the structure of clinical workload more flexible to allow summarized higher level
data with lower level detail
- Make data entry and retrieval easier and more effective
- The use of language and structures must be dealt with before integration of person based
system
o Clinical teams developed evidence based order set templates.
o Decision support tools help identify the appropriate evidence based set.
o Edit template for individualized Plan
- Page 16.580
Most current knowledge provide the basis for each care
o Deviation from order set to refine the evidence base and added to clinical knowledge
3) Implementation of systems
5 key factors for successful implementation
a. Having organizational leadership, commitment and vision
b. Improving clinical processes and patient care
c. Involving MD in the design and modification of the system
d. Maintaining or improving clinical Productivity
e. Building Momentum and support amongst MD
4) Study of the effects of the system
a. Decrease the time for documentation
b. Improve the quality and relevance of data in the record
c. Increase the proportion of nursing time spent in direct patient care
d. Verbal orders eliminated
e. Communication enhanced among providers
f. Prospective review of clinical data provided concurrent assessment of patient
progress
g. Reduced cost
h. Improving patient safety and outcomes
i. Detect adverse events
- Page 16.582
- Patient care changing in 2 ways
1) Legacy systems for charge capture replaced by support and improve clinical practice, plus
send clinical data to locations for practice, management, and research
2) Support discipline separately shift to integrated, interdisciplinary concepts of care
Research to:
1) Develop structured clinical languages, standards, and data models
2) Develop innovative systems
3) Determine more effective and efficient ways to implement systems
4) Investigate the effects of changing information resources on the processes of care and the
functions of organization
Chapter 15: Public Health Informatics and The Health Information
Infrastructure - Page 15.538
- Public Health Informatics: The systematic application of information and computer science and
technology to public health practice, research and learning
- Distinguished by
1) Focus on population
2) Orientation to preventions
3) Government context
Public Health Focus
1) Public Health focuses on the health of community
2) Treatment to community such as public disclosure of the disease status of the individual
patient
3) Environment factors (e.g. air quality)
4) Large Emphasis on the prevention of disease
The nature of a given intervention is determined by
1) Cost
2) Expediency
3) Social acceptability of intervention at any potentially effective point
- E.g. waste water treatment, housing codes
- Page 15.539
Public health 3 core functions:
1) Assessment: Monitoring and tracking the health status of populations including identifying
and controlling disease outbreaks and epidemics
2) Policy development: utilizes the results of assessment activities and etiologic research in
concert with local values and culture to recommend interventions and public policies that
improve health status
- E.g. seat belt laws
- Use Web to promote health
3) Assurance: the duty of public health agencies to assure their constituents that services
necessary to achieve agreed upon goals are provided
- Services, such as medical care, might be provided by the public health agency or by requiring
other public or private entities to provide the services
- Page 15.540
- To assure that all members of the community have adequate access to needed services
- Include clean water, safe food supply
- The core function of assessment rely heavily on public health surveillance: Ongoing collection,
analysis, interpretation, and dissemination of data on health condition (e.g. breast cancer) and
threats to health (e.g. smoking prevalence)
- Surveillance data useful both in short term (infectious disease) and longer term (determining the
leading causes of premature death)
- Data are collected for the purpose of action to guide public health response (e.g. out break
investigation) or to help direct public health policy (e.g. obesity)
10 essential services of Public Health
1) Monitor health status of individuals in community to identify community health problem
2) Diagnose community health problems
3) Inform community to health issues
4) Mobilize community partnership in identifying common health problem
5) Develop policy to improve health
6) Enforce laws to protect public health
7) Link individuals to community providers
8) Ensure competent workforce for the provision of public health services
9) Research new insight to community health problems
10) Evaluate the effectiveness of personal and population based health services in a community
- Page 15.541
- Epidemiology: The study of the prevalence and determinants of disability and disease in
population
- There is no uniform national routine reporting for most diseases, disabilities, risk factors, or
prevention activities in USA
Public health information systems features:
1) Optimized for retrieval from very large record databases
2) Able to quickly cross tabulate
3) Study secular trends
4) Look for patterns
- E.g. CDC HIV/AIDS reporting system
- National Notifying Disease Surveillance system
- MMWR: Morbidity and Mortality Weekly Report
- CDC WONDER system
- Data from periodic surveys from stand alone information systems provide periodic estimates of
incidences
- No finer detail than a region
- Many are patient self reported
- Page 15.542
- Disease registries that track the incidence of certain conditions
- Focus on one topic for specific time periods
- Only a small fraction of medical care
- Difficult to Access
- They all rely on special data collection
- Rarely seamlessly linked to ongoing clinical information systems
- Each clinician data is reentered
- Duplicate data entry
- Data being shallow, delayed
- Subject to error
- National Electronic Disease Surveillance System (NEDSS) promote the use of data and
information system standards for integrated surveillance system
- Need system that can tell us about individuals and the world in which these individuals live
- Public health and clinical informaticians work together
Public health informatics example
- Immunization Registries
- Confidential, population based, computerized information systems that contain data about
children and vaccinations
- Need because of:
1) Scattering of immunization records among multiple providers
2) Complex immunization schedule
3) Reduced in the incidence of disease: sense of complacency
- The All Kids Count (AKC) program
- Standards needed
- CDC worked closely with Health Level 7 standards development organization to define HL7
messages
- Developed an immunization registry development guide
- Page 15.545
Challenges in 4 areas
1) Interdisciplinary Communications
- HL7 must accurately present and enable the complex concepts and processes that underlie the
specific business process required
- Need clear communications among health providers
- Lack of shared vocabulary
- Anxiety and concerns of new information system
- Power shifts
- Need to identify an interlocution familiar with both IT and public health
- Page 15.546
2) Organizational and Collaborative Issues
- Large number and wide variety of partners
- Need a tool to minimize their time and expense for registry data entry
- All stakeholders need to be represented in the decision making processes, guided by a mutually
acceptable governance mechanism
- Legislative and regulatory issues must be considered
- Issues of confidentiality data transmission and liability are critical
- HIPAA compliant
3) Funding and Sustainability
- Need Development of a business case that shows the anticipated costs and benefits of registry
- Helpful in prioritizing requirements
4) System design
- Issues include
a. Data acquisition
b. Database organization
c. Identification and matching individuals
d. Generating recommendations
e. Access to data
- Page 15.549
Most of the community Health information networks are not successful because
1) Standards and technology not yet ready for cost-effective community based electronic
health information exchange
2) Focus on availability of aggregated health information for secondary users (policy
development) rather than for direct provision of patient care
3) No sense of urgency
4) No substantial funds available
- Attention to cost, quality, and error issues in the health care system
- Building a national health Information Infrastructure (NHII)
- Vision and Benefits of NHII
- Anytime, anywhere health care information at the point of care
- To create a distributed system
- Patient information collected at each care site
- Various existing electronic records would be located, collected, integrated, and immediately
delivered
- Receive reminders of the most recent clinical guidelines and research results
Benefits of NHII
- Error reduction
- Remind MD about recommended action at point of care
- Notifications of actions
- Page 15.550
- Warnings about planned treatments
- Improve safety and reduce cost
- Improve dissemination of new research results
- Improve the efficiency of clinical trials
- Data collection during administering the protocol
- Analyzing de-identified aggregate data
- Page 15.551
- Early detection of patterns of disease
- Early detection of possible bioterrorism
- NHII would allow electronic reporting of both relevant clinical events and lab results to public
health
- Page 15.552
- Substantially reduce HC cost
Barriers and challenges to NHII
1) Protecting the confidentiality of EMR
2) Misalignment of financial incentives in the HC system
- Benefits do not accrue equally across all segments of the system
- E.g. for individual HC providers
- Page 15.553
- Most of the NHII financial benefits accrues to payers of care
- ROI is uncertain
- Legal barriers prevent transfer of funds from those who benefits from HIT to those with no
returns
3) Lack of interoperable electronic medical record systems that provide easy transfer of
records from one place to another
Approaches to accelerating NHII process
1) Establishing standards
- The CHI recommended 5 key standards in 2003:
o HL7 v 2.x
o LOINC
o DICOM
o IEEE 1073
o NCPDP SCRIPT
- Government licensed the SNOMED
- Joint efforts of IOM and HL7 to develop definition of EHR
- Development of a formal interchange format standard (IFS)
- Page 15.554
- Need representation of guideline recommendations
- Development of an effective guideline interchange standard
- Page 15.555
2) Promoting collaboration
- E.g. eHealth Initiative
- National alliance for Health Information technology (NAHIT)
3) Demonstration projects in communities
- Provisions of seed funding
- Reasons:
a. Existing models of HII are based in local communities
b. System may not be developed in larger scale
c. Increase size increase complexity and risk of failure
- The need for trust to overcome confidentiality concern
- Need legal agreements and policy changes
- Page 15.556
d. Keep implementation more reasonable
e. Customized approach for each community
f. Encourage communities to utilize national standards
- Page 15.557
- Seed funding is essential in the development of LHII systems
- Need to channel existing funds to LHII
4) Measure to evaluate progress
- Measures should be:
a. Sufficiently sensitive so that their values change at a reasonable rate
b. Comprehensive enough to reflect activities that impact most of the stakeholders
c. Meaningful to policy makers
d. Periodic determinations of the current values of the measures should be easy
e. The totality of all measures must reflect the desired end state
- E.g Aggregate measures:
o % if population covered by an LHII
o Settings of care
- Progress in implementation of EHR
o Health care functions performed using support e.g. CPOE
o Progress in semantic encoding of EHR
o Usage of EHR
- Page 15.559
Vision for application of NHII to homeland security
o Early detection of bioterrorism and the response to such an event
o Clinical information relevant to public health reported electronically in near real time
o Specific disease reported to dynamically adjust in response to an actual incident
o NHII provide more effective medical care resource management in response to events
o Automatic reporting of all available resources for proper allocation
- Page 15.559
Challenge of HII in homeland security
1) Participation from a wide range of public and private organizations
2) Incompatible information systems
3) Lack standardize terminology and exchange
4) Policy conflicts e.g. Access to information
5) Ambiguous governance structure
6) Need new types of system
7) Need for interdisciplinary communication
Chapter 13: Management of Information in HC organization - Page 13.476
- HC professionals comprise a heterogeneous group with diverse objectives and information
requirements
- The purpose of HCIS is to manage the information that HC professionals need to perform their
jobs effectively and efficiently
- Page 13.477
- Shifting of financial risks from third party payers to the hospitals
- Rapid consolidation of providers into Integrated Delivery Networks (IDN)
- Largest HCIS focus on the automation of specific functions within hospitals
- Then to support ancillary department
- Aka Hospital Information Systems (HIS)
- Lack of connectivity among systems
- Don’t know
o locations of patients
o What kinds of care provided
o Clinical results of the care
- Page 13.478
- Clinical Information System for order entry and results communications
- Ambulatory medical record systems (AMRS) and practice management systems (PMS) to
support large outpatient clinics and MD offices
- Isolation of these systems
- The development of the interface engine
- The creation of HL7, a process for standardizing the content of data messages sent from one
system to another
- Challenge of sharing data among different IS
- Need new system to address the complexity and cost of interface
- Interface engine to capture multiple systems called Enterprise Application Integration
- Creation of HL7 to move data among systems
- Page 13.479
- Mergers among organizations
Forms Integrated Delivery Networks (IDNs)
Goals of IDNs
1) To reduce costs and increase revenues by negotiating with 3rd party payers
- Some affiliated with HMO
2) Cut cost by achieving economies of scale by consolidating administrative and financial
functions and coming clinical services
- Page 13.480
- Expertise gained from managing an inpatient driven organization (e.g. hospital) does not
translate easily to the successful management of other organizational activities or to other
hospitals
- Page 13.481
- For MD: HCIS to present patient specific data to caregivers
- For Administrator: daily operation and management of organization
- Page 13.482
HCO Operational Information needs
1) Operational Requirements
- HC workers require detailed and up to date factual information to perform the daily tasks
- Organizing data for prompt and easy access
2) Planning requirements
- Information to make short term and long term decisions about patient care
- Choose resources to provide high quality care at competitive price
3) Communication requirements
4) Documentation and reporting requirements
- Patient health status and treatment history
- Legal document
- Page 13.484
HCO Integration need
1) Data Integration
- Traditionally, clinical and administrative data are handled separately
- Redundant data entry and data maintenance
- Inconsistent data
- Sharing of data among operating units becomes critical and problematic
- Transfer of patient
- Medication info in ER
2) Process Integration
- Information systems must mesh smoothly with operational workflow and human organizational
system
- Mechanism for information management aimed at integrating operations among entities must
address migration from legacy systems and migration from legacy work processes to new
consistent policies and processes across entities
- Page 13.486
- Real value comes only when underlying work processes are changed to take advantage of the
new information technology
- To reduce costs
- Increase productivity
- Improve service levels
- Most clinical sites must adapt their own processes to those embedded in the system
- Workflow became part of the organization’s culture
- HCO must change quickly to meet the continually evolving requirements of today’s HC
environment
- Security and confidentiality requirements
- HIPAA
- HCO must strike a balance between restricting information access and ensuring the
accountability of the users of patient information
HCO must adapt a 3-pronged approach to securing information
1) Designate a security officer and develop uniform security and confidentiality policy
2) Train employees so they understand the appropriate uses of patient identifiable information
and the consequences of violations
3) Use electronic tools such as access controls and information audit trails
- Page 13.487
Benefits of HCIS
1) Cost reduction
- Reduction in labor requirements
- Reduced waste
- More efficient management of supplies and other inventories
- Efficient scheduling of resources
- Eliminate inadvertent ordering of duplicate tests
- Reduce the cost of storing, retrieving, and transporting charts
2) Productivity Enhancements
- Enable staff to manage a larger variety of tasks and data
3) Quality and Service improvements
- Improved accuracy and completeness of documentation
- Reduction in time MD spent in documenting (increase time spent with patients)
- Fewer drug errors and quicker response to adverse events
- Improve provider to provider communication
- Telemedicine expand reach and improve delivery of specialist care to rural areas
- Page 13.489
4) Competitive advantage
- Return on security investments
- The benefits of avoiding security breaches
- Ability to influence clinical practices by reducing large unnecessary variations in medical
practices
- To improve patients outcome
- To reduce costs
- Provide MD with access to information on best practices based on the latest available clinical
evidence
- Ability to satisfy external organization
- Identify patients for enrolling in research trials
5) Regulatory Compliance
- E.g. FDA mandates the use of barcodes on all drugs
- E.g. HIPAA specify the required content and format for electronic data transactions
- Page 13.490
Managing IS in a Changing HC environment
- Grand challenge to design and implement an HCIS that is sufficiently flexible and adaptable to
meet the changing needs of the organization
- Avoid implementing a system that is obsolete functionally or technologically before it becomes
operational
- Page 13.491
1) Changing technologies:
- Decide whether to upgrade individual systems and interfaces to newer products or to migrate to
a more integrated systems environment
- The end point is not fixed
2) Changing culture:
- MD aimed at reducing variation in care
- Focus in the cost of care
- Assume responsibility for sick and wellness of people
- Work as member of a collaborative patient care team
- Time allocated for a patient visit in the ambulatory setting is decreasing
- Pay for performance incentive to reward desired work practice
- Become information managers
- Page 13.492
3) Changing process
- Critical to the success of evolving HCO: Developing a new vision of how HC will be delivered and
managed, designing processes, and implementing supporting information systems
- Poor IS implementation can institutionalize bad processes
- Work process redesign is essential
- Handling of people and process issues: one of the most critical success factors for HCO
4) Management and Governance
- A formal governance structure with representation from all major constituents provides a
critical forum for direction settings, prioritization, and resource allocation across an HCO
- Page 13.494
Functions and components of a HCIS
1) Patient Management and Billing
- Use computer based master Patient Index (MPI) to store patient identification information,
basic demographic data that are acquired during the patient registration process
- MPI can be integrated in the registration of an ambulatory care system or elevated to an
enterprise master care patient index (EMPI) across several facilities
- Within the hospital setting, the census is maintained by the admission discharge transfer (ADT)
module , which updates the census whenever a patient is admitted to the hospital, discharged
from the hospital, or transferred to a new bed
- Serve as a reference base of billing functions
- Provide a common reference base for use by other patient care settings
- Scheduling System
- Patient tracking application to monitor movement in process
- Page 13.495
- Integrated Delivery networks ensure unique patient identification through implementation of an
enterprise EMPI
2) Departmental Management
- Ancillary departmental systems support the information needs of individual clinical departments
within an HCO
- 2 purposes
a. Perform many dedicated tasks required for departmental operations
- E.g. Generating specimen collection list
- Printing medication labels
b. Contribute major data components to online patient records
- E.g. lab results, digital images
3) Care delivery and clinical documentation
- Automated order entry
- Results reporting
- To communicate with ancillary department electronically
- Computer based patient record system (CPRs)
- Page 13.496
- Developed diagnostic specific clinical pathways that identify clinical goals
- Document actual vs expected outcomes
- Closed loop medication systems
- Every task is recorded in the HCIS
- Use of hand held devices
- Page 13.497
4) Clinical Decision Support
- Directly assist clinical personnel in data interpretation and decision making
- To monitor patients and issue alerts
- To make diagnostic suggestions
- To provide limited therapeutic advice
- To provide information on medication cost
- Integration with other functions such as CPE
- Clinical event monitors
5) Financial and Resource management
- Management of payroll, human resources
- Complicated by the many reimbursement requirements of government and third party payer
- Page 13.498
- Burden of financial risks for care shifted from 3rd party payers to providers
- The growth of managed care has added more complexity
- Goal to reduce the cost per unit service
- Maintain the health of members using health resources effectively
- Provider Profiling system: Track each provider’s resource utilization (e.g. cost of drugs
prescribed, diagnostic test ordered) compared with severity adjusted outcomes of that
provider’s patients (such as their rate of hospital readmission and mortality by diagnosis_
Contract Management Systems
- Estimating costs associated with potential managed care contracts and comparing actual with
expected payment based on contract terms
- Handle Patient Triage and medical management functions, direct patients to appropriate health
services
Historical evolution of HIS
1) Central and Mainframe based System
- Focus on Patient management and billing functions
- E.g. Technician Medical Information system (TMIS)
- E.g. Center of Clinical Computing System (CCC)
2) 1970’s Departmental Systems
- One or a few machines are dedicated to processing specific functional tasks within the
organization
- E.g. Lab systems
- E.g. DHCP: Distributed Hospital Computer Program for the VA hospitals
- Success due to focus on the clinical environment
- They do not have to conform to the general standards of the overall system, so they can be
changed to accommodate the special needs of specific areas
- E.g. system for ICU different from Radiology
- Modification of department system is simpler
- Increased difficulty in integrating data and communicating among modules
- Page 13.501
- By 1980’s HCIS based on network communications technology were being developed
- As distributed systems, connected through electronic networks
- UCSF
- Advantage: Individual department have a great deal of flexibility in choosing HW and SW that
suits their needs
- Disadvantage: the distribution of information processing and responsibility for data among
systems makes the task of data integration communication, and security difficult
- Individual departments encode data values in ways that are incompatible with the definitions
chosen by other areas of organization
- Need to define data standards
- Today, distributed architecture is the norm
- PC based universal workstations are the norm
- Page 13.502
3) Integrated Systems from Single Vendors
- Turn key systems
- Departmental Functions integrated into application suites, developed by a single vendor
- Page 13.503
- In the 1990’s, IT strategies based on the use of integrated systems from vendors historically
focused on smaller hospitals
- Architecture for a changing environment
- Not feasible for an IDN to replace all legacy systems with new common systems
- A single architecture is unlikely to suffice for all
Lessons learned:
1) A strategy for data preservation by providing access to data and implementing an approach
for standardizing the meaning of those data
2) IDNs and HCOs should separate 3 conceptual layers to allow greater flexibility
- Page 13.505
a. Data Layer
- Plan must include access to data for applications and a method to ensure data collected across
business units are consistent and comparable
- Security and confidentiality safeguards
i. For clinical data, need for both real time operations and retrospective data
analysis
- Clinical data repository: serves the need of patient care and day to day operations
- Data warehouse: serves longer term business and clinical needs such as contract management
and outcomes evaluation
ii. Keep patient information comparable
- Need to identify the patient
- The enterprise master patient index (EMPI) as the name authority
- An index of patients names and identification numbers used by all information systems in the
IDN that stored a patient registry
- A consistent approach must be developed for naming data elements and defining their values
- E.g. know which patient is allergic to penicillin
- Which patients for new prevention services
- Develop their own internal vocabulary standards, or terminology authority
- Page 13.506
- CPMC separates the storage and retrieval of data from the meaning of the terms in the database
using medical entities dictionary (MED)
- Develop a set of terminology services, 3 categories:
1. Link the data within HCO legacy databases before copied to CDR
2. Re-registering all terms used by new applications and linking them
to external authoritative vocabulary terms
- E.g. in the UMLS metathesaurus
3. Providing real time help in selecting the appropriate term to
describe a clinical situation
b. Business Logic Layer
- Separate the workflow or business logic from the database will enable more natural migration
of systems
- Enable organization to change workflow
c. User Interface Layer
- Most subject to frequent changes
- Use thin clients to reduce costs
- Driven by the internet
- Handheld device for accessing schedule
- Modified hand writing capabilities
- Voice entry devices
- The design of the display and the nature of the input devices should not be so tied to the
application that change and modification are difficult
- Page 13.507
Forces that will shape the future of HIS
1) Changing Organizational Landscape
- Serve target patient population
- De consolidation of IDN
- IDN’s changes fast, but technology may require years to build
- Operational Budget continue to shrink
- New investments translate to increased operating costs
2) Technological Changes
a. Emergence of powerful microprocessor and cost of storage media
b. Expanding availability of Internet access within organizations and to patients homes
c. Design of modern software based on code standards such as XML
d. Distributed HC capabilities
- Page 13.508
3) Societal Change
- MD Spending less time with patients and more time with administrative and regulatory
concerns
- Consumers have access to more health information
- Changing model of care with changing economic incentives
- Greater focus on wellness and preventative and lifelong care
- Align economic incentives with wellness
- More information can be stored efficiently on movable media
Chapter 14: Consumer Health Informatics and Telehealth - Page 14.511
- Complexity and collaboration characterize HC in the early 21st century
- Page 14.512
- Patient: The most underused resource in the HC delivery systems
- Electronic communication have the potential to improve care by reducing the cost and delays
associated with travel
- Telemedicine involves the use of modern IT, to deliver health services to remote patients and to
facilitate information exchange between PCP and specialist at some distance away from each
other
- Telemedicine emphasizes the distance, especially the provision of care to remote patients
- Biomedical Informatics emphasizes method for handling the information
- Telemedicine: videoconferencing between patients and providers
- Telehealth include both traditional medicine and interactions with automated systems or
information sources
- Page 14.513
- Telehealth derived from traditional patient care
- Consumer health informatics (CHI) derived from the self health movement of the 1970’s
- Patient participation takes many forms
- Shared decision making
- Self care
- Collaborative practices
- By self monitoring
- By evaluating and choosing therapeutic strategies from a set of acceptable alternatives
- By implementing the therapy
- By evaluating the effects
- Page 14.514
- In the distributed managed care models of health care, consumers serve as their own case
managers
- Information tools such as EMR, provide an integrated record and communication service
Historical Example
- The use of “Leper Bells” in Roman
- The Australian Royal Flying Doctor Service (RFDS)
- Standardized Medical Chest
- Page 14.515
- In 1964 Teleconsultation in for remote mental health facility
- 1970 to 1980: Video based telehealth
- Military teleradiology displayed in 1991
- Correctional Telehealth became much more common
- Hampered by 2 problems
1) High cost
2) Poor image quality
- Page 14.516
- By 2000: Improvements in image compression made it possible to transmit low resolution, full
motion video over standard phone lines
- Telehome care
Engaging consumers in Health Care
- Early 20th century: US Federal Children’s Bureau served as the major source of health
information for the public
- In 1950: use mainframe computer system as a health assessment tool
- In 1980: Computers use for computer assisted learning program
- Page 14.517
Categories of telehealth and consumer health informatics
o Information Resources
o Messaging
o Telephone
o Remote Monitoring
o Remote Interpretation
o Video Conferencing
o Telepresence
- Separation into Synchronous (real time) or asynchronous (store and forward system)
- Synchronous Telehealth
o E.g. Telephony
o Chat groups
o Telepresence
- Asynchronous Telehealth involves the preparation of dataset at one site and then sent to a
remote recipient
o E.g. Remote interpretation
o Teleradiology
- Page 14.518
- Consumer health informatics resources provide substantive and procedural knowledge about
health problems and promising interventions
CHI resources provide patients with condition specific and disease specific information
about the problems they face
- Page 14.519
- Originated from 2 perspectives
1) Professional developed CHI resources are those developed by HC MD and their
organizations
- HC organization develop information resources as a service to the patient population that they
treat
- E.g. Kaiser P Health Facts
- Mayo Health Adviser
- Health Wise
- Health desk
- Page 14.520
2) Self help perspective: address daily living concerns and lifestyles issues along with content
by established medical authorities
- E.g. Fred Hutchinson Cancer research center
- EHR’s relevance to individual patient’s group
- Provide direct patient access to clinical record
- Page 14.522
- Provide opportunities for individuals to connect with other people who share similar concerns
and with their HC providers
- Network based CHI resources generally provide both static information about health problems
and management and specialized health communication utilities
- Facilitate communication between the patient and the health plan
- Web Q&A services
- Discussion groups and chat rooms
- Patient to patient communication are the most actively used aspects
- Timely access to staged, tailored, informational services may have the greatest positive benefit
- Page 14.523
- Text messaging: popular mode of communication between patients and provider
- Use of web based messaging solutions called Personal Clinical electronic communications
- E.g. Medem System
- Patient log onto a secure web site to send or view messages
- Page 14.523
- Up to 25% of all primary care encounters occur via the telephone
- The value of telephone followup for chronic conditions
- Managed care companies setup large telephone triage center
- Remote monitoring: capture of clinically relevant data in the patient’s home and the subsequent
transmission of the data to central locations for review
- Page 14.524
Remote monitoring focuses on management, rather than on diagnosis
o Collection of discrete measurements
o Measurement of the parameter and transmission of data are separate events
o By data entry or by direct data transfer
o E.g. glucometer
o Spirometer
o Daily weights
- Limited use because
1) Question of efficacy
2) Who will review the data
3) Money
- Remote interpretation: Stored and forward telehealth that involves the capture of images atone
site and transmission to another site for interpretation
- E.g. radiographs (Tele radiology)
- PACS: Picture Archiving and Communication Systems
- Teleophthalmology
- Page 14.525
- Video based telehealth
- Most use a hub and spoke topology
- Led to scheduling problem
- Availability of relevant clinical information
- Require stable data stream
o Telementor
o Video Camera in ambulances
- Page 14.526
1) Telepsychiatry
- Diagnosis based primarily on observing and talking to patient
2) Correctional telehealth
3) Home Telehealth
- Use of Plain old telephone service (POTS) connection
- With data ports for connection of peripheral devices
- Page 14.528
- 2 categories
a. Telehome care: Telehealth equivalent of home nursing care
- Focus on recovery from a specific disease
b. Management of chronic disease
- Involves a longer duration of care and less frequent interactions
- Focus on patient education
- E.g. IDEA TEL using home telemedicine Units (HTU)
- Page 14.529
- Telepresence: allow MD to act on remote situations
- E.g. Telesurgery
- Requires Teleoperation of robotic surgical instruments and accurate force feedback
- Need extremely low network latency
- Allow MD to make remote video rounds
- Page 14.530
Challenges and Future directions
1) Quality of information and content credentialing
- Need certification by recognized bodies
- Disadvantages of credentialing
a. Challenge to ensure that every information element is tested and evaluated fully
b. Leaves control of the authority for HC information in the hand of the traditional care
providers
c. Contradictory to HC consumerism that empowers consumers to make choices
consistent with their own world view
Use 6 criteria to evaluate
i. Credibility
ii. Content
iii. Disclosure: Purpose of site
iv. Links
v. Design
vi. Interactivity
vii. Caveats
- Credibility: includes the source, currency, relevance/utility, and editorial review process for the
information
- Content: Must be accurate and complete, and an appropriate disclaimer provided
- Disclosure: includes informing the user of the purpose of the site, as well as any profiling or
collection of information associated with using the site
- Links: Evaluated according to selection, architecture, content, and back linkages
- Design: encompasses accessibility, logical organization (navigability), and internal search
capability
- Interactivity: Includes feedback mechanisms, means for exchange of information among users,
and tailoring of information to individual needs
- Caveat: clarification of whether site function is to market products and services or is a primary
information content provider
2) Challenges to Using internet for consumer health and telehealth applications
- Security of transmission
- Ensuring every citizens access to the internet
- Quality and integrity of many devices
3) Licensure and Economics in Telehealth
- Medical licensure in US is state based
- Law to require face to face encounter before any electronic care
4) Reimbursement
- Page 14.533
5) Roles of Health Professionals in CHI: 3 Roles:
a. Sources for content
b. Guidance in moderating public electronic discussion groups and responding to
patient’s electronic messages
c. Information brokers and interpreters for patients
- Page 14.534
- Future problems
- MD to become information brokers and to devise new ways to ensure that patients are
prepared to participate in the clinical interactions
- Changes in the sequential paradigm
- MD filled with alerts and results everyday
- MD need to monitor many processes at once
- Share decision making with patients
- Involve patients early and continuously
- Integrate patient preferences
- Break down of role, geographic, and social barriers
Chapter 4: Cognitive Science and Biomedical Informatics - Page 4.134
- Cognitive Science is the multidisciplinary domain of inquiry devoted to the study of cognition
and its role in intelligent agency e.g.
1) Cognitive Psychology
2) Artificial Intelligence
3) Neuro Science
4) Linguistics
5) Anthropology
6) Philosophy
- Cognitive Science focus on fundamental aspects of cognition (e.g. Attention memory, early
language requisition)
- Applied research focus on the development of useful cognitive artifacts
- Cognitive artifacts are human made devices that
1) Extends people’s abilities in perceiving objects
2) Encoding and retrieving information from memory
3) Problem Solving
- Aligned closely with disciplines of human computer interaction (HCI) and human factors
- Page 4.135
- Cognitive Science and Biomedical Informatics
Cognitive Science can provide insight into
1) Principles of system usability and learnability
2) The process of medical judgment and decision making
3) The training of HC professionals
4) The study of collaboration in the workplace
3 Areas:
1) Basic cognitive scientific research and theories
2) Research in the area of medical cognition
3) Applied cognitive research in biomedical informatics
Cognitive research, theories, and methods can contribute to applications in informatics in
1) Seed basic research findings that can illuminate dimension of design (e.g. attention and
memory aspects of the visual system)
2) Provide an explanatory vocabulary for characterizing how individuals process and
communicate health information (e.g. medical cognition pertaining to doctor – patient
interaction
- Page 4.136
3) Present an analytic framework for identifying problems and modeling certain kind of user
interactions
4) Develop and refine predictive tools (GOMS method of analysis)
5) Provide rich descriptive accounts of clinicians employing technologies in the context of work
6) Furnish a generative approach for novel designs and productive applied research programs
in informatics
- Behaviorism: A framework for analyzing and modifying behavior
Logical positivism: All statements are either
1) Analytic (true by logical deduction)
2) Verifiable by observation
3) Meaningless
- Page 4.137
- Empiricism is the view that experience is the only source of knowledge
- Behavioral theories of learning emphasize the correspondence between environment stimuli
and the response emitted
- Page 4.138
- Problem solving can be construed as search in a problem space
- A problem space has an initial state, a goal state, and a set of operators
- Operators are any move that transform a given state to a successor state
- The search process involves finding a solution strategy that will minimize # of steps
- Page 4.139
Protocol analysis: The most common method of data analysis
- A class of techniques for representing verbal think aloud protocol
- Think aloud protocols are the most common source of data used in the study of problem solving
- Subjects were instructed to verbalize their thoughts
- Page 4.140
- Human Information processing
- Much of human cognition can be characterized as a series of operations or computations on
mental representations
- Mental representations are internal cognitive states that have certain correspondence with the
external world
- Page 4.141
E.g. 2 interdependent principles by which we can characterize cognitive systems are:
1) Architectural theories that endeavor to provide a unified theory for all aspects of cognition
- Categorize regularities of human information processing system
a. Structural regularities (the relationship between perceptual and memory systems
and memory capacity limitations)
b. Processing regularities (e.g. processing speed, selective attention, or problem
solving strategies)
2) Distinction among different kinds of knowledge that is necessary to attain competency in a
given domain
- Page 4.142
- The ACT-R cognitive architecture developed by Anderson (1983)
- To model a wide range of cognitive phenomena pertaining to language, learning, and memory
and perception and motor skills
A cognitive layer with 2 long term memory modules
1) A procedural memory related to how to execute particular actions (e.g. moving a mouse)
and perform various activities
2) Declarative memory (conceptual knowledge) that contains concepts ( medical findings and
disorders)
- The perceptual motor layer consists of
1) Effector (Motor and Speech)
2) Receptor (vision and auditory) which interacts with the world and the cognitive layer
- Page 4.143
Human memory:
1) Long term memory (LTM): A repository of all knowledge
2) Working memory (WM) : resources needed to maintain information active during cognitive
activity)
- E.g. text comprehension
- WM is limited to 5 to 10 chunks of information
- Cognitive load: An excess of information that competes for few cognitive resources creating a
burden on WM
- Page 4.143
- Cognitive Architectures
- E.g. SOAR initiated GOMS(Goals, Operators, Methods) and selective Rules
o For describing a task of very fine level of granularity (e.g. the keystrokes) and the user’s
knowledge of how to perform the task
- Page 4.144
- Humans actively construct and interpret information from their environment
- Representation enables us to remember, reconstruct, and transform objects absent in space
from the initial encodings
- The represented world: the cognitive representation of individuals that are the subject of
investigation
- The representing world: The means by which we attempt to capture the dimensions of
representation selected by inquiry
- The function of the representing world is to present information about the represented world
- Proposition are a form of representing that captures the essence of an idea, without explicit
reference to linguistic content
- E.g. “Hello”, “Hey”, has identical context
- Page 4.145
- Propositional knowledge can be expressed using a predicate calculus formalism
o A detailed way to characterize the information subjects understood from reading a text,
based on their summary or explanations
- Comprehension involves an interaction between what the text conveys and the schematic in
LTM
- The text information is called the textbase
- The situational model is constituted by the textbase representation and the domain specific
knowledge
- Schematic represents a higher level kind of knowledge structure
- Data structures for representing the generic categories of concepts stored in memory
- Page 4.146
- When a person interprets information, the schema serves as a filter for distinguishing relevant
and irrelevant information
- E.g. Schema for MI may contain the findings for chest pain, seating, etc.
- Page 4.147
- Mental Images: Captures perceptual information recovered from the environment
- Mental Model: Describes how individuals form internal models of systems
- Page 4.148
- Conceptual Knowledge refers to one’s understanding of domain specific concepts
- Procedural knowledge is a kind of knowing related to how to perform various activities
- Often modeled as production rule: A condition action rule that states “If conditions are satisfied”
then execute the specified action
- Page 4.149
- Factual knowledge
- Involves knowing a fact without any indepth understanding
- Developing medical ontologies for use in knowledge based system:
E.g. An epistemological framework of
o Observation level
o Finding Level
o Fact level
o Diagnosis Level
o System complex layer
- Page 4.150
- To characterize knowledge used for medical understanding and problem solving, and for
differentiating the levels at which medical knowledge may be organized
- Study of expertise
- Superior recognition ability is not a function of superior memory, but is a result of an enhanced
ability to recognize typical situations
- Chunking: A chunk is any stimulus that has become familiar from repeated exposure and is
subsequently stored in memory as a single unit
- Novices worked backwards from the unknown problem solution
Experts worked forward
- Findings
- Page 4.151
1) Experts are capable for perceiving large chunks of information in their domain
2) Fast at processing and deployment of different skills
3) Superior short term and long term memory in their domain of expertise
4) Represent problems in their domain at a deeper level
5) Spend more time assessing the problem prior to solving it
6) Experts differs substantially
- A domain expert processes an extensive accessible knowledge base that is organized for use in
practice and is tuned to the particular problem at hand
- Page 4.152
Different levels of expertise
1) Lay person: only common sense
2) Beginner: Has the pre-requisite knowledge assumed by the domain
3) Novice: Lay person or beginner
4) Intermediate: above beginner but below subexpert level
5) SubExpert: With general knowledge but inadequate specialized knowledge of the domain
6) Expert: With specialized knowledge of the domain
- Human Learning requires the arduous process of continually learning, re-learning, and exercising
new knowledge, punctuated by periods of decrease in mastery and declines in performance,
which may be necessary for learning to take place
- Intermediate effect in many tasks
- Page 4.153
- During intermediate effects, a reorganization of knowledge and skills takes place, characterized
by shifts in perspectives or a realignment or creation of goals
- People at intermediate level generate a great deal of irrelevant information
- Page 4.154
- Intermediate not under time pressure processes too much irrelevant information
- Novice lack the knowledge to do much search
- Page 4.155
Hypothetico-deductive process
- 4 stages:
1) Cue acquisition
2) Hypothesis generation
3) Cue Interpretation
4) Hypothesis Evaluation
- Novice’s Knowledge was described to be “classically centered”
- Experts memory of disease model was found to be extensively cross-referenced with a rich
network of connections among diseases
- Page 4.156
- Experts use data driven reasoning: which depends on the MD processing a highly organized
knowledge base about the patient disease
- Novice use hypothesis driven reasoning resulting in very complex reasoning patterns
- Pure data driven reasoning is only successful in constrained situation where one’s knowledge of
a problem can result in a complete chain of inferences
- Data driven reasoning: Error prone
- Hypothesis driven reasoning is slower
- Used when domain knowledge is inadequate or the problem is complex
- Page 4.157
- Data driven reasoning breaks down in case complexity include the presence of “loose ends”:
Information unaccounted for
- Use of data driven reasoning may lead to cognitive heavy load
- Page 4.158
- Experts: the ability to abstract the underlying principles of a problem
- Page 4.159
- Human Computer Interaction: A cognitive engineering approach
- Difficult Interfaces result in deep learning curves and structural in-efficiencies in task
performance
- More money spent on usability evaluation
Usability methods:
o Observations
o Focus groups
o Surveys
o Experiments
- Page 4.160
Usability includes 5 attributes:
1) Learnability: System easy to learn
2) Efficiency: Experienced user can attain a high level of productivity
3) Memorability: Features easy to retain once learn
4) Error: Minimize error
5) Satisfaction
- Heuristic Evaluation: A usability inspection method in which the system is evaluated on the
basis of a small set of well tested design principles: e.g.
1) Visibility of system status
2) User control and freedom
3) Consistency and standards
4) Flexibility and efficiency of use
- Emphasize simplicity and functionality
- System failed because the capabilities of the system are not readily usefully deployed to
improve human performance
- HCI is a multifaceted discipline devote to the study and practice of usability
- Page 4.161
-
- Cognitive Engineering: A cyclical pattern of interaction with a system
Norman’s 7 stage model of action
1) Goal: E.g. retrieving a patient’s health record
2) Formation of an intention: Retrieve the record outline
3) Specification of an action sequence: e.g. Logging on to the system -> Entering patient’s
health record #
4) Executing an action
5) Perception of a change in system state
6) System response to be interpreted
7) Evaluated
- Page 4.162
2 primary points for break down
1) Gulf of execution reflects the difference between the goals and the intention of the use and
the kinds of actions enabled by the system
- E.g. need to press enter at the end of an action
2) The Gulf of evaluation reflects the degree to which the user can interpret the state of the
system and determine how well their expectations have been met
- Due to differences in the designer’s model and the user’s mental model
- Page 4.163
- Norman’s Theory of action has given rise for the need for sound design principles
- Informed a range of cognitive task analytic usability evaluation methods
- E.g. cognitive walkthrough
- Cognitive task analysis: Analyses on both the information processing demands of a task and the
kinds of domain specific knowledge required to perform it
- Cognitive Walkthrough (CW): Characterize the cognitive processes of user’s performing a task
- Identifying sequences of actions needed to complete a task
- Aim to determine whether the user’s background knowledge and the cues generated are likely
to be sufficient to produce the correct goal-action sequence required to perform a task
- Performed by analyst walking through the sequence of actions necessary to achieve a goal
- Page 4.161
- CW Analysis includes:
1) Goals: e.g. Open Excel
o Subgoals: Locate icon on desktop
o Actions: Double click
2) System response: Change in screen
3) Attempt to discern potential problems
- Page 4.162
- Minimize # of actions to complete a task
- Minimize screen transitions
- External representation are vital sources of knowledge
- Display serves as input to cognitive system for further processing
- Page 4.163
- External representation plays a critical role in enhancing cognition and intelligent behavior
- Representational Effect: Different representations of a common abstract structure can have
significant effect on reasoning and decision making
- E.g. Arabic numerals more efficient than roman numerals
- Display can reduce the amount of time spent searching for critical information
- Page 4.166
- E.g. A Matrix table for taking medicine
- Instructions can be embodied in a range of external representations from text to list of
procedures to diagram
- Involves both quantitative and qualitative reasoning
- Page 4.168
- The law of proximity: Visual entities that are close together are perceptually grouped
- The law of symmetry: Symmetric objects are more readily perceived
- Information visualization: The use of computer supported, interactive, visual representations of
abstract data to amplify cognition
- Page 4.169
5 major classes of representation types:
1) List
2) Table
3) Graph
4) Icon
5) Generated text
- Each has distinct measurement properties
Criteria for evaluating the efficacy of a representation
1) Latency: the amount of time it takes a user to answer a question based on information in
the representation
2) Accuracy
3) Compactness: the relative amount of display space required for representation
- Page 4.170
6 Ways external representations can amplify cognition:
1) Increasing the memory
2) Processing information available to the users (off loading cognitive work to a display)
3) Reducing the search for information (grouping data strategically)
4) Using visual presentations to enhance the detection of patterns
5) Using perceptual attention mechanisms for monitoring (e.g. drawing attentions to events
that require immediate actions)
6) Encoding information in a manipulative medium (users can select different possible views to
highlight variables of interest)
- Distributed cognition and electronic health records
- Display can have a central role in controlling interaction in graphical user interfaces
- Distributed Cognition (DC) represents a shift in the study of cognition from being the sole
property of the individual to being “stretched” across groups, material artifacts, and cultures
- In the distributed approach: Cognition is viewed as a process of coordinating distributed internal
(i.e. knowledge) and external (e.g. visual displays, manuals) representation
DC has 2 central points of inquiry:
1) Emphasizes the inherently social and collaborative nature of cognition (e.g. doctors, nurses,
and technical support in prenatal care jointly contributing to a decision process
2) The mediating effects of technology or other artifacts in cognition
- Page 4.171
- A more moderate perspective in which an individual’s mental representation and external
representations are both given recognition as instrumental tools in cognition
Mediating Role of Technology:
1) The effects with technology
- Concerned with the changes in performance displayed by users while equipped with technology
- E.g. when using an HIS, alleviate cognitive load and focus on higher order thinking skills
- Page 4.172
2) The effects of technology: refers to the enduring changes in general cognitive capacities
(knowledge and skills) as a consequence of interaction with technology
- Page 4.173
- Screen driven strategy can enhance performance by reducing the cognitive load and allow MD
to allocate more cognitive resources toward testing hypotheses and rendering decisions
- For novice users: Get irrelevant findings and pursued incorrect hypotheses
- The subject become too reliant on technology and had difficulty imposing his own set of working
hypotheses
- Changes in knowledge organization
- EHR contained more information relevant to the diagnostic hypotheses
- The structure and content of information corresponds to the structured representation of the
particular medium
- E.g. EHR more information about the patient’s past medical history
- Paper based records appears to better preserve the integrity of the time course of evolution of
the patient problem
- Page 4.174
- The enduring effect of technology even in the absence of the particular system
- Structure and content of paper based records bear close resemblance to the organization of
information in the EHR
- Experts are much less bound to the order and sequence of presented information on the EHR
screen
- IT can mediate cognition and produce enduring changes in how one performs a task
- Well designed artifacts could reduce the need for users to remember large amount of
information
- Poorly designed artifacts increased the knowledge demands on the user and the burden of WM
- Affordance: Attributes of objects that enable individuals to know how to use them
- E.g. the affordance of a door handle
- Page 4.175
- Most cognitive tasks have an internal and external component
- The problem solving process involves coordinating information from these representations to
produce new information
- DC paradigm: can be used to understand how properties of objects on the screen (e.g. links,
buttons) can serve as external representation and reduce cognitive load
- The configuration of resources (e.g. very long menus, complexity configured displays) placed
unnecessary heaving cognitive demands on the user)
- Redistribution and reconfiguration of resources might yield guiding principles and design
solutions in the development of complex interactive systems
The use of clinical practice guidelines desired results
1) Adoption of best practices
2) Decreased variability
- Page 4.176
- Failure because of lack of integration into workflow
- Mismatch between guidelines and the MD opinion of what should be done for a specific patient
in a particular clinical setting
- CPG can be semantically complex
- The process of interpretation consists of 2 components
1) What the text itself say
2) What each person interprets from the text
- Any 2 interpretations should be equivalent in terms of the information that is inferred
- Guidelines should allow for flexibility in interpretation
- Page 4.177
Guideline Development Process
1) Generation of the paper guidelines at authoring institutions, such as the American College of
Physicians ( ACP), which involves the evidence based consensus development
2) Transition of paper guidelines into an algorithm form to be easily used by MD
3) Translate the paper based guidelines into computer based representations
4) Implementation of the computer based representation in clinical situations
5) End user’s interpretation of the guideline as it is represented in the guideline applications in
specific clinical settings
- Page 4.178
- Comprehension is an essential step in design, problem solving, and decision making
- Page 4.179
Process of translation from Internal Representations into natural and computer represent
able languages
- The final product of discourse processing consists of what the text “says” and whatever the prior
knowledge and experiences the users “attach” to the text
- Representation varies from reader to reader as prior knowledge differ
- The techniques of propositional and semantic analysis identify ambiguous areas in the text that
lead to misunderstanding
- Page 4.180
- Propositional analysis is a formal method for investigating representation of meaning in memory
- A creation of a semantic network that provides a picture of the whole conceptual
representation at any desired level of detail
- Semantic network are graphs consisting of a non-empty set of nodes and asset of links
(structure) connecting such nodes (content)
- Convey 2 types of information: Conceptual (the concepts load) and structural (how the concepts
related to one another)
- Page 4.191
- Md need the availability of evidence
- Also need is the format of guideline presentation
- Representation can have great influence on decisions in medicine
- Experts use guidelines as reminders during the problem solving process
- Nonexperts used guidelines as an aid to knowledge reorganization
- Guidelines help PCP to separate relevant information from irrelevant ones
- Page 4.192
- GLIF3: Guideline Interchange Format: A sharable guideline representation format
o Allow for a formal specifications of medical concepts and data
- Guidelines must be flexible enough to accommodate different level of expertise of user
- Fine balance between flexibility and the inclusion of detail necessary for informational and
computational equivalence
Chapter 9: Imaging and Structural Informatics - Page 9.344
- Imaging generation: the process of generating the images and converting them to digital form
- Image manipulation: uses pre-processing and post-processing methods to enhance, visualize or
analyze the images
- Image management: methods for storing, transmitting, displaying, and organizing images
- Image integration is the combination of images with other information needed for
interpretation
- Page 9.343
- Purpose to extract information about the structure of the body
- Overlaps structural informatics: the study of methods for representing, organizing, and
managing diverse sources of information about the physical organization of the body
- Neuro informatics: Neuro imaging
- Digital Imaging represented by a 2 dimensional array of numbers (a bit map)
- Each element of the array represents the intensity of a small area if the picture called a pixel
- For the image of a volume, a 3D array of numbers are required, each element represents a voxel
- Page 9.346
Imaging Parameters
1) Spatial Resolution: related to the sharpness of the images: related to the number of pixel
per image area
2) Contrast resolution: a measure of the ability to distinguish small differences in intensity
- # of bits per pixel
3) Temporal resolution: a measure of the time needed to create an image
Perfect Image Modality
1) High Spatial, Contrast, and Temporal Resolution
2) Low in Cost
3) Portable
4) Free of Risk
5) Painless
6) Noninvasive
7) Non-ionizing Radiation
8) Depict Physiologic Functions
- Page 9.347
4 areas of development
1) Energy Source
a. Light
b. X Ray
- Film based Radiography
- Computed Radiography (CR)
- X Ray is differentially absorbed by various body tissues
- The X Rays produce shadows on the radiographic film
- The resultant shadowgraph is a super position of all the structures traversed by each beam
- Use film and Fluoroscopic screens
- Pros:
o High Spatial resolution
o Medium cost
o Can be generated in real time
o Produced using portable instruments
- Cons
o Poor contrast resolution
o Use of ionizing radiation
o Inability to depict physiologic functions
c. Ultrasound
d. Nuclear Magnetic Resonance
- Images from magnetism grew out of nuclear magnetic resonance (NMR) spectroscopy
- Creation of images from NMR signals known as MRI had to await the time development of
computer based reconstruction images
- Page 9.349
2) Reconstruction methods
- Contrast Radiography: use of radiopaque contrast material to highlight the areas of interest
- Angiography
- In 1970: The first commercially viable computed tomography (CT) scanner
- CT mathematically reconstructs an image from X-Ray Attenuation values from multiple angles
- Page 9.350
3) Higher Dimensionality
- CT, PET, MRI: can be made 3D by acquisitions of a series of closely spaced parallel slices
4) Contrast Agents
- Histologic Staining Agents
- Magnetic Contrast Agents
- New and emerging Structure Image Method
- Page 9.352
- Charge Coupled Device (CCD) cameras convert existing film based equipment to units that can
produce images in digital form
- Magnetic Resonance Arteriography (MRA)
- Diffusion Tensor Imaging (DTI)
- Page 9.353
- Ultrasound in 3D using volume or surface based rendering techniques
- Confocal Microscope uses electronic focusing to produce a 3D voxel array of microscopic
specimen
- Electron Tomography
- Molecular Imaging
- Page 9.354
- 2 D Image Processing
- Page 9.355
- Involves the transformation of input images into output images or into some abstract
representation of the contents of the input images
4 Image Procession Steps
1) Global Processing: Computations on the entire image, without regards to specific local
content
- To Enhance an image
- E.g. Gray scale windowing of CT images
2) Segmentation: Extraction of regions of interest (ROIs) from the overall image
- Edge Detection techniques are used
- Or by Region Detection Techniques
3) Feature Detection: Process of extracting useful parameters from the segmented regions
- May be used as input into an automated classification procedure
4) Classification: Determines the type of object found
Mathematical models use to aid in the performance of image analysis subtasks
- Primary uses are for:
1) Image enhancement
2) Screening
3) Quantitation
- Page 9.356
Image Processing Techniques Example
1) Image Enhancement
- E.g. CT windowing
- Unsharp Masking: Blurred image subtract from original image
- Histogram equalization: spread the image gray levels throughout the visible range
- Temporal Subtraction: Subtract reference image from later images
- E.g. Digital Subtraction Angiography (DSA): Background image subtracted from image with
contrast material
2) Screening determine whether an image should be flagged for careful review
3) Quantitation: Characterize meaningful regions of interest
3D Image Processing
- Additional Informatics Issues
- Page 9.358
1) Registration:
- Each voxel represents the image intensity in a small volume of space
- A voxel must be accurately registered in the 3D volume (voxel registration)
- Separately acquired image volumes from the same subject must be registered with each other
(volume registration)
- Voxel registration: Almost all CT and MR manufacturers consoles contain same form of 3D
registration and visualization capabilities
- To align the sections with each other, embed a set of thin rods in the tissue, to manually indicate
the location of these fiducials in each section
- Transform each slice so the corresponding fiducials line up in 3D
- Page 9.359
Volume Registration
- Align separate image volumes from the same subject
- Intrasubject alignment
- Different image modalities provide complementary information: Therefore acquire more than
one kind of image volume on the same individual
- E.g. combining MRV and MRA with MRI
- Primary problem to solve in multimodality image fusion is volume registration
- If patient moves, other registration methods are needed: Image can be aligned to templates of
the same modalities that are already aligned
- Page 9.360
2) Spatial representation of Anatomy
- The 3D Image can be visualized using volume rendering techniques
- Which project a 2D image directly from a 3D voxel array
- Image volume is processed in order to extract an explicit spatial representation of anatomy
- Segmentation and Reconstruction Technology
a. Reconstruction from serial sections
- Semiautomatic contour tracing followed by semiautomatic tiling
- Most common method for reconstruction from serial sections
- Reconstruction from serial sections is the method of choice for extracting microscopic 3D Brain
anatomy
- Page 9.361
b. Region Based and edge based segmentation
c. Model and Knowledge Based segmentation
- Use of deformation Model
d. Combine Methods
- Page 9.363
3) Symbolic Representation of Anatomy
- Attach labels to the structures
- Terms from symbolic qualitative models (ontologies) of anatomic concepts and relationships to
support systems that manipulate and retrieve segmented structures
- If anatomic ontologies are linked to other ontologies of physiology and methodology they can
provide knowledge about the meaning of various images
- E.g. Nomina Anatomica
- Terminologia Anatomica
- NeuroNames
- Page 9.364
- Foundational Model of Anatomy (FMA)
- A comprehensive symbolic description of the structural organization of the body, including
anatomical concepts, their preferred names and synonyms, definitions, attributes, and
relationships
- Page 9.365
- FMA useful for symbolically organizing and integrating biomedical information
4) Atlases: Spatial representation of anatomy are combined with symbolic representations to
form digital atlases
- A digital Atlas is generally created from a single individual, which serves as a “canonical”
instance of the species:
- E.g. 2D: the Digital Anatomist Interactive Atlases
- 3D: Digital Anatomist Dynamic Scene Generator (DSG)
- E.g. Voxelman: Voxel in the visible human hand is labeled with the name of an anatomic
structure in a generalized voxel model
- Page 9.367
- Talairach Atlas: Most widely used human brain atlas
5) Anatomic Variation
- Develop method for relating the anatomy of multiple brains
- 2 major approaches
a. Warping to a template atlas
- Warp an individual target brain to a single brain chosen as a template
- How to register 2 brains as closely as possible
- Page 9.368
i. Volume based warping
2 approaches
1. Intensity Based
2. Landmark based
ii. Surface based warping: To register 2 cortical surfaces
Common reconfiguration techniques
o Inflation
o Expansion to a sphere
o Flatening
- Page 9.369
b. Population Based Atlas
- 3 Main Methods
i. Density Based
ii. Label Based
iii. Deformation Based
6) Characterization of Anatomy
- Page 9.370
a. Qualitative Classification: Group individual structures into various classes based or
perceived patterns
b. Quantitative Classification: Morphometrics or computational neuroanatomy
- Page 9.373
Functional Imaging
- By observing changes of structure over time
- Page 9.374
1) Image Based Functional Brain Mapping
- From scanners that generate low resolution volume arrays depicting spatial localized activation
- E.g. PET, MRS
- E.g. Functional Magnetic Resonance Imaging (FMRI)
2) Nonimage based Functional Mapping
- Data mapped to anatomy, and then displayed as functional overlaps on anatomic images
- E.g. Cortical Stimulation Mapping (CSM)
- For localizing functional areas in the exposed cortex at the time of neurosurgery
Chapter 11: Evaluation and Technology Assessment - Page 11.402
- Evaluation: describes a wide range of data collection activities, designed to answer casual and
focused questions
- Technology assessment: any process of examining and reporting properties of a medical
technology used in health care such as safety, efficacy, feasibility and indication for use, cost,
and cost effectiveness, as well as, social, economic, and ethical consequences, whether intended
or unintended
- Medical technology: Techniques, drugs, equipment, and procedures used by health care
professionals in delivering medical care to individuals, and the systems within which such care is
delivered
- Page 11.403
5 Reasons to study clinical Information resources
1) Promotional: to encourage the use of information resources in medicine
2) Scholarly: Develop clinical information resources using computer based tools
3) Pragmatic: Need to know which techniques are more effective
4) Ethical: ensure it is safe
5) Medico legal: reduce the risk of liability
- Page 11.404
Stakeholders in Evaluation Studies
1) Developers
2) Users
3) Patients
4) People responsible for purchasing and maintain the system
Challenges of study design and conduct
- Page 11.405
1) The complexity of medicine and HC Delivery
- Influence 3 aspects of HC system
a. HC systems’ structure
- The space it takes up
- The equipment available
- Financial resources required
- Staff
b. Processes
- # and appropriateness of diagnoses
- Investigations and therapies administered
- Page 11.406
c. Outcomes
- Quality of Life
- Complication of procedure
- Length of survival
- Information for one group maybe of little benefit to other
- Rigorous proof of Safety and effectiveness is required in evaluation
- Medicine is a complex domain
- Patients diseases are different
- No quantifiable changes in patient outcomes after the Introduction of many information
resources
- Processes of decision making are complex
- Information resources may intervene in the process of medical decision making
- Lack of gold standards in medicine
- Wide variation in consensus opinion
- Page 11.409
2) The Complexity of computer based information resources
- NP Hard: To test a program rigorously requires the application of every possible combination of
all possible input data in all possible order
- N! experiments: N is the # of input data items
- Computers adapt to one institution may be difficult to conduct at a different time at another
location
- Rapid evolution of HW and SW
- Need to be familiar with the components of the information resource
- Page 11.410
3) The Complexity of Study methods
- Studies of medical information resources require analytical tools from the behavioral and social
sciences, statistics, and other fields
- Study require test materials and users in short supply
- The manner in which case data are abstracted and presented to users
- Many questions can arise
- Page 11.411
Full Range of what can be studied
1) The clinical need the resource is intended to address
2) The process used to develop the resource
3) The resource’s intrinsic structure
4) The functions that the resource carries out
5) The resource’s effect on users, patients, and other aspects of the clinical environment
Evaluating focus changes in different aspects
1) The clinical status quo without the resource
2) The Skills of the development teams and the methodologies employed
3) Specifications, flowcharts without running the program
4) The resource performs when it is used
5) The influence on users
- Page 11.412
Factors that characterize an evaluation study
1) Focus of the Study: e.g. the status quo, design process
2) Study Setting: e.g. In a lab setting, in a clinical setting
- Page 11.412
3) Clinical data employed: e.g. simulated data, data from real patient
4) User of the resource: E.g. developers, end users
5) Decisions affected by the use of the resource: e.g. simulated decisions, real decisions
- Approaches to study design
- Page 11.414
Anatomy of all studies
1) Evaluation begin with a process of negotiation to identify the questions, for the study
2) Understanding of how the evaluation will be conducted and an initial expression of the
questions
3) Investigations: Collection of data to address these questions
4) Mechanism for reporting back to the individual who need it
5) Format of the report inline with the stipulations of the contract
- Typologies of Evaluation methods
- Best by Ernest House
- Page 11.415
- Each approach is linked to an underlying philosophical model
- 4 Objectivist Approaches
- 4 Subjectivist Approaches
Major Premises underlying the objectivist approach: Logical Positivist
1) Attributes of interest are properties of the resource under study
- The worth of an information resource can be measured with all observations yielding the same
result
2) Rational persons can agree on what attributes of a resource are important to be measured
and what resources are the most desirable
3) Numerical measurements is superior to a verbal description
4) Falsification: Never possible to fully prove a hypothesis
5) Can demonstrate that a resource is superior to what it replaced
Intuitionist Pluralist Subjectivist approaches to evaluation
1) Different observers have different conclusions
2) Merit and worth explore in context
- Page 11.416
- The value of a resource emerges as it functions in a particular environment
3) Individuals hold different perspectives
- Document the ways in which they disagree
4) Verbal description can be illuminating
5) Evaluation: an exercise in argument
Approaches to Evaluation
1) Comparison Based Approach: employs experiments and quasi experiments
- Comparison with a control condition, a placebo, or a contrasting resources
- Comparison based on a small number of outcome variables assessed in all groups
2) Objective Based: Determine whether a resource meets its designer’s objectives
- E.g. System’s advise available within 15 minutes of patient first seen
3) Decision facilitation: resolve issues important to developers and administrators so that
these individuals can make decisions about the future of the resource
- E.g. A systematic study of alternative formats for computer generated advisories
4) Goal Free: Evaluators purposely blinded to the intended effects and identify all effects of the
resource, intended or not
5) Quasi-Legal: establishes a mock trial, to judge a resource
- E.g. treatment of sickle cell
6) Art Criticism: A respected critic works with the resource
- Critic writes a review highlighting the benefits and shortcomings of the resource
- E.g. Software review
- Page 11.418
7) Professional Review: Panels of experienced peers who spend several days in the
environment where the resource is installed
- Result is a report
8) Responsive Illuminative: Represent the viewpoints of both users of the resource and people
who are an otherwise significant part of the clinical environment where the resource
operates
- Goal is understanding or illumination, rather than judgment
- Investigators immerse themselves in the environment where the resource is operational
- Page 11.419
Stages of technology Assessment
1) Emphasizes Technical Characteristics
- E.g. The response time of an IS to a query
2) Efficacy or effectiveness of a device
- E.g. Clinical trials of IS
- Use Process Measures
- The degree of MD compliance with computer generated reminders
3) Directly evaluate effectiveness via health and economic outcomes
- Most comprehensive technology assessments
- Page 11.419
- E.g. Computer-based reminder system for breast cancer screening would examine changes in
mortality or morbidity from breast cancer
- Evaluation the cost of system
- Uncommon in medical informatics because outcomes are infrequent
- Page 11.420
Stages of Evaluations
1) First stage: Technical Characteristics Evaluation of the design and development processes of
a clinical information resource or of the structure of the resources
- The software engineering of the resource
- How the resource integrate with other systems
2) Second stage: Clinical Efficacy
- Evaluate the functions of the resource
- Clinical Trials
- Effect of IS on the process of care
3) Third Stage: Comprehensiveness Clinical Effectiveness, Economic, and Social Outcomes
- Establish a technology’s cost effectiveness
- Comprehensive assessment of health and economic outcomes
- E.g. Study in terms of $ per Quality: Adjusted Life Year (QALY) saved
- Depends on the purpose of the study and on the cost and feasibility of measuring the outcome
- Challenging in Medical Informatics
- Structure and terminology of comparative studies
- Creates a contrasting set of conditions to compare the effects of one with those of another
- Goal is to attribute cause and effect or to answer questions raised by other kinds of studies
- Page 11.421
- Participants: Entities about which data are collected
- Variables are specific characteristics of the participants that either are measured or are self
evident characteristics of participants
- Page 11.422
- Dependent variables: Form a subset of variables in the study that captures the outcome of
interest
- Aka outcome variables
- Dependent variables will be computed , as an average over a number of tasks
- E.g. measure MD’s diagnostic performance
- Independent Variables: included in a study to explain the measured values of the dependent
variables
- E.g. whether a computer system is available, to support certain clinical trials
- The dependent variable is some type of performance measure that involves concern about
reliability (precision) and Validity (accuracy) of measurement
- Accuracy = degree of closeness to the true value
- Precision = Reproducibility
- Page 11.423
- Measurement is the process of assigning a value corresponding to the presence of a specific
attribute or a specific object
- Result in:
1) The assignment of a numerical score representing the extent to which the attribute of
interest is presence in the object
2) Assignment of an object to a specific category
- People planning studies are faced first with the task of deciding what to measure and then with
developing their own measurement methods
- Process of measurements
- Investigator applies Instruments / Observation to Object to infer Value of Attributes
- Page 11.424
- Measurement Studies: determine how accurately an attribute of interest can be measured in a
population of objects
- All observers will agree on the result of the measurement
- Demonstrative studies: Directly address questions of substantive and practical concern
- E.g. Measurement of Study explored how accurately the speed of an information resource can
be measured
- Demonstration study explore whether a particular resource has sufficient speed to meet the
needs of busy clinicians
- Investigators should know their measurement method will be adequate BEFORE they collect
data for their studies
- Control strategies in comparative studies
- Subjects who complete tasks that are not affected by the intervention of interest
- Page 11.425
- Study example
- Intervention: Installation of a reminder system
- Subjects: MDs
- Terms: Patients cared for by MD
- Dependent Variables: Includes:
o MD ordering of antibiotics
o Rate of past operative infections averaged across the patients
Types of studies
- Page 11.425
1) Descriptive (Uncontrolled) Studies
- Install system
- Training
- Make measurements
- No Independent variables
- Hard to interpret figures without comparison
2) Historically Controlled experiments
- Aka Before / After study
- Investigator makes baseline measurements before the system in installed
- Make the same measurements after the system is in use
- Independent Variable: Time
- 2 levels: Before and After installation
- Many other factors may have changed in the interim to affect the study
- Page 11.426
- Should add either internal or external controls or both
- Internal control: measure likely to be affected by any non specific changes happening in the
local environment, but unaffected by the intervention
- External control can be the same as in the target environment but in a similar external setting
e.g. in another hospital
3) Simultaneous Controls
- Make our outcome measurements in doctor and patients who are not influenced by the
reminder system but who are subject to the other changes taking place in the environment
- Take measurements before and during the intervention
- A parallel group comparative study with simultaneous control
- Dependent variable: Post operative infection rate
- Independent variable: Time and Group
- There may be unmeasured differences
- Page 11.427
4) Simultaneous Randomized controls
- Randomize the assignment of subjects to control or intervention groups
- Any difference that is statistically significant (e.g. p value < 0.05) can be attributed to the
reminder
- Page 11.428
- Internal Validity: Confident in the conclusion drawn from the specific circumstances of the
experiment
- External Validity: Conclusions can be generalized from the specific setting to a broader range of
setting that other people will encounter
4 possible outcomes
1) The resource was truly effective, our study shows effective
2) Resource was ineffective, our study show it is not effective
3) Resource is effective, our study did not show effective
- False negative
- Type II error
- System effect
4) Resource is ineffective, our study suggest it was effective
- False Positive
- Type I error
- P < 0.05 : 5 % of making a type I error
Threats to Internal Validity
1) Assessment bias: All persons do not allow their own feelings and beliefs about an
information resource to bias the results
2) Allocation Bias: Allocate easier cases to the information resource group (allocation bias) or
avoid recruiting a particular easy (or difficult) case to the study
3) Hawthorne Study: Tendency of human to improve their performance if they are known
being studied
- The study itself caused the increases
4) Checklist effect: Improvement due to more complete and better structured data collection
- To control, collect the same data in the same way in the control and in study group
5) Placebo effect: Patients feel good about receiving attention and potentially useful
medication
- Page 11.430
Cost effectiveness and Cost benefit studies
- To assess quantitatively the benefits obtained from a health intervention relative to the cost of
the intervention
- Cost effectiveness analysis: Analyst expresses the health benefits in units of health outcomes
and the cost in dollars
- Incremental Cost Effectiveness ratio: the appropriate estimate of the relative value of the
intervention
- (C(B) – C(A)) / (LE(B) – LE(A))
- C(B) = Cost of intervention “B”
- LE(B) = Life expectancy with intervention B
- Cost Benefit Analysis values all benefits and costs in dollars
Determine whether the benefit ($) is larger than the cost ($)
- Steps:
1) Define the problem
- Objective and perspective of analysis
- Alternative intervention
- Determine the decision context
- What decision does the analyst need to make
- Identify the objective of the study: e.g. reduce hospital costs or all costs
- Determine the perspective of the study: e.g. hospital setting
- Identify the alternatives
2) Identify and analyze the benefits
- Decide how to measure the health benefit of the system
- Should compare only with other similar interventions
- Choose a more comprehensive measure of health outcome
- Estimate the costs with the old system vs the new system
3) Identify and analyze the costs
- Page 11.432
- Direct cost: Value of all goods, services, that are required to produce an intervention, including
resources consumed because of future consequences of the intervention
- E.g. change in use of HC resources, non HC resources, patient time, costs of drugs, tests,
facilities
- Direct non HC costs include other services required
- E.g. Transportation costs
- Family members
- Time Spent by patient
- Do not need to include these costs
- Productivity costs are those costs that accrue because of changes in productivity due to illness
or death
4) Perform discounting
- Enables the analyst to account for time preference in the analysis: expenditures or benefits that
occur immediately
- Calculate the net present value of health outcomes and costs
- The cost effective threshold reflects the value judgement of the decision makers about the
maximum value a year of life saved
5) Analyze uncertainty
6) Address ethical questions
7) Interpret the results
- Page 11.433
- Rationale for Subjectivist studies
- Represents the viewpoints of users of the resource
- Goal is Illumination
- Methods used derive from Ethography: Investigators immerse themselves physically in the
environment where the resource is, and they collect data through observations
- The designs are not rigidly pre-determined
- Page 11.434
Methodology of formal Systems analysis
o A process of information gathering
o Heavily reliant on interviews
o A cyclic, iterative process
o An overly structured approach can mis-portray
Differences between objectivist and subjectivist approaches
1) Subjectivist studies are emergent in design
- Objectivist studies:
o Begins with a set of hypothesis
o A plan to address them
o Page 11.433
o Subjectivist studies begin with orienting issues that stimulate the early stages of
investigation
Important questions emerge
Investigation is willing to adjust future aspects of the study
Investigators are incrementalist
Change their plan from day to day and have a high tolerance for uncertainty
2) A naturalistic Orientation: A reluctance to manipulate the setting of the study: the
environment
- They cannot alter the environment to study it
- Employ quantitative data for descriptive purposes
- E.g. when MD and nurse both use a clinical system to enter orders, their experiences with the
system offer a natural basis for comparison
3) End product is a report written in narrative prose
- No technical understanding is required
- Results are accessible to a broad community
Natural history of a subjectivist study
1) Negotiation of the ground rules of the study: Between the study team and the people
committing the study
- The aim of study
- The kinds of methods used
- Access to various sources of information
- The format for final report
- Aims in a set of orienting questions
- Results expressed in a memorandum of understanding
- Page 11.436
2) Immersion into the system
- Introduction to conversation
- Silent presence at meetings
3) Iterative loop: a cycle of:
o Data collections: Interviews
o Analysis and reflection
o Member checking: sharing of investigations emerging thoughts with the participants
o Reorganization: Revised agenda for data collection in the next cycle
o Backward steps are natural
4) Preliminary Report: Share with users to obtain check on the validity of the findings
- Use of anonymous quotations
5) Final report: distributed in the original memorandum of understanding
- Meet the “investigator” session
Data Collection and Data Analysis Methods
1) Observation
- Participant observation: Observer becomes a member of the work team
- Detached observer
2) Interviews:
- Formal interviews: both investigators and interviewees are aware that the answers to the
questions are being recorded
- Unstructured interview: no predetermined questions
- Semistructured interview: Flexible on the order of topics
- Structured interview: Same question in the same word in the same order
- Informal interview
3) Document and Artifact analysis
- These artifacts do not change
- Unobtrusive measures: record accrue as part of the routine use of the resource
4) Anything else that seems useful
- E.g. clinical chart reviews, questionnaires
5) Analysis of subjectivist data
- Analysis conducted systematically
- Look for themes emerging from different sources
- Conclusions derive credibility from a process of “triangulation”: information from independent
source generate the same theme or point to the same conclusion
- Page 11.439
The Mindset of evaluation and technology assessment
1) Tailor the study to the problem and key stakeholders questions
2) Collect the data that will be useful to make decisions: No limit to the questions ask or data
collected
3) Look for intended and unintended effects
4) Study the resources while it is under development and after it is installed
- 2 kinds of decisions
a. Formative decisions are made as a result of studies undertaken while a resource is
under development
- They affect the resource before it can go online
b. Summative decisions: are made after a resource is installed and deals with how
effectively the resource performs in that environment
5) Study the resource in the lab and in the field: In vitro studies vs in vivo studies
6) Go Beyond the developer’s point of view: Get close to the end user
7) Take the environment into account: goodness of fit with its environment
8) Let the key issue emerge overtime: Evaluation studies are dynamic
9) Be methodologically catholic and eclectic
- Derive overall approaches, study designs, and data collection methods form the questions to be
explored rather than to bring pre-determined methods or instruments to a study
Chapter 20: Clinical Decision Support System - Page 20.698
Types of Decisions
- Page 20.699
1) Diagnosis: The pathophysiological explanation
2) Diagnostic Process: which questions to ask, tasks to order
- What data are needed
3) Management Decisions
Requirements for decision making
1) Accurate data
2) Pertinent knowledge
3) Appropriate problem solving skills
- The data about a case must be adequate, but they must not be excessive
- Important: the quality of available data
- Page 20.700
- Decision makers must have broad knowledge
- Their knowledge must be current
- Good problem solving skills are important
- CDSS is any computer programs designed to help HC professionals to make clinical decisions
3 types of decision support functions
1) Tools for information management
- HC information systems
- HC information retrieval system
- Specialized knowledge management workstations
- Interpretation is left to the MD
- Page 20.701
2) Tools for focusing attention
- Flag for abnormal values
- Provide list of explanations for these abnormalities
3) Tools for providing patient specific recommendations
- Custom tailored assessments based on sets of patient specific data
- E.g. Dxplain: Diagnostic assistant
- Page 20.702
Leeds Abdominal Pain System
- F.T. DeDombal
- Used sensitivity, specificity, and disease prevalence data for various signs, symptoms, and test
results to calculate, using Bayes’ Theorem, the probability of 7 possible explanations for acute
abdominal pain
- Assumption
1) Conditional Independence of the findings for the various diagnosis
2) Mutual Exclusivity of the 7 Diagnoses
- Page 20.703
- The system never achieve the same degree of diagnostic accuracy in other settings
- Different training and different cultures
- Different between findings and diagnoses in different patient populations
- MYCIN: De-emphasized diagnosis to concentrate on appropriate management of patients who
have infections
- Knowledge of infectious disease was represented as production rules
- A production rules is a conditional statement that relates observations to associated inferences
that can be drawn
- Early exploration of methods for capturing and supplying ill-structure knowledge to solve
medical problems
- HELP: an integrate hospital information system developed in LDS hospital: uses PAL language
- Ability to generate alert when abnormalities are noted
- Adds a monitoring program
- Adds a mechanism for storing decision logic in sectors
- In 1990: Arden Syntax: For writing rules that relate specific patient situations to appropriate
actions
- Each Decision rule or Help Sector, is called a Medical Knowledge Module (MLM)
- Integration of decision support with other functions can heighten a program’s acceptance and
encourage its use
- Page 20.707
Change in acceptance of CSS due to:
1) The emergence of PC, WWW, and easy interface
2) Computer systems must meld with work practices
3) Large amount of information needed
4) Practice cost effective, evidence based medicine
Characterize CDSS by:
- Page 20.708
1) System Functions:
- 2 categories:
a. Determine what is true about a patient (the current diagnosis)
b. What to do for the patient
- Many assist MD with both activities
- e.g. diagnostic programs help MD to decide what information would be most useful in
narrowing the differential diagnosis
- Must balance the cost and benefits of action
2) The model for giving advice: Passive vs Active
- Md must recognize when advice would be useful and make an explicit effort to assess the
program
- Challenge is to avoid generating excessive numbers of warnings for minor problems
- Page 20.709
3) Style of Communication
a. Consulting Model: the program serves as an advisor
o Accepting patient data, asking questions, and generating advice for the user
b. Critiquing Model: Acts as a sounding board for the user’s own ideas, expressing
agreement or suggesting reasonable alternatives
o E.g. ATTENDING for anesthetic selection
4) Underlying decision making process
- Problem specific flowcharts
- Use of simple printed copies of algorithm
- Use detailed algorithmic logic
- Mathematical Modeling
- Pattern recognition
- Page 20.710
- Bayesian Diagnostic programs
- Use of belief networks to develop more expensive Bayesian system in which conditional
dependencies can be modified explicitly
- Use methods of decision analysis
o Adds to Bayesian reasoning the idea of explicit decisions and of utilities associated with
various outcomes that could occur
o Influence Diagrams
o Use of Artificial Neural Networks (ANN)
o When the correct diagnosis may depend on interactions among the findings that are
difficult to predict
- Page 20.711
Development of a knowledge based system:
o encodes concepts derived from experts in a field
o In a knowledge base
o To provide analysis and advice
- The knowledge base encodes a non-numeric, qualitative model of how inferences are related to
reach abstract conclusions about a case
5) Human computer interaction
- CDSS should present interfaces to their users that are intuitive
- Users can predict in advance of their actions
- Fail if ask users to move to a separate workstation
- Follow complex startup procedures
- Lengthy interactions
- Page 20.712
- Manual re entry of information
- Solutions:
o Link computers together
o Use of wireless networks
o Write on tablet
o Speech recognition
Barriers to CDSS:
1) Acquisition and Validation of patient data
a. Data Entry problems
- Need combination of speech and graphics
- Coupled with integrated data management environment to prevent redundant data entry
b. Lack standardized ways of expressing most clinical situation
2) Modeling of Medical Knowledge
- Page 20.713
- Decide what clinical distinction and patient data are relevant
- Identifying the concepts for the decision making task
- Ascertaining a problem solving strategy
- Need a model of both the required problem solving behavior and the clinical knowledge
3) Elicitation of medical knowledge
- Page 20.714
- Methods to facilitate the development and maintenance of medical knowledge bases
- E.g. OPAL for cancer chemotherapy advisor ONCOCIN
- Use of meta-tool to generate automatically a specific purpose knowledge elicitation tool based
on the model e.g. Protégé
- Page 20.715
4) Representation of and reasoning about medical knowledge
- Challenge to represent anatomical knowledge and performing spatial reasoning by computer
- Challenge in knowing how to use what is known
- Aka good clinical judgment
5) Validation of system performance
- Gold standard difficult to define
6) Integration of decision support tools
- Page 20.716
Examples of CDSS
- Diagnosis CDSS:
o Quick Medical Reference (QMR) supports diagnostic problem solving in general internal
medicine
o DXPlain: Web based diagnostic system
o EON: Guideline based decision support system
o Provide therapeutic recommendation for treatment in accordance with pre-defined
protocol
- Page 20.720
- QMR’s use as an electronic reference is far more important than its use as a consultation
program
- Page 20.722
Patient Management: Guideline Based Architectures and the EON system
- Clinical Practice Guidelines are a powerful method for standardizing and uniform improvement
of quality of care
- Systematically developed guideline to assist practitioners and patients decisions about
appropriate HC for specific clinical circumstances
Several tasks that can benefit from automation
1) Specification and maintenance of clinical guidelines
2) Retrieval of guidelines appropriate to each patient
3) Runtime applications of the guideline
4) Retrospective assessment of the quality of the application of the guideline
- Page 20.723
- Developers encode the knowledge for raising context sensitive alerts as situation action rules
- A rule interpreter processes such rules
- Page 20.724
- E.g. ProForms language
- E.g. the Guideline Interchange Format (GLIF)
- Page 20.725
- EON: a second generation knowledge based system that aids practitioners in the care of patients
who are being treated in accordance with protocols and clinical practice guidelines
- EON Cannot run by itself
- Page 20.726
Major components of EON
1) Problem Solver
a. Determining the treatment that should be given at a particular time
b. Determine any protocol eligible
2) Knowledge bases
3) Database mediator: Serves as the conduit between all the problem solvers in EON and the
database for patient data
- Includes a problem solver: the RESUME system
- The components can be mixed and matched to create different decision support functionalities
- E.g. T-HELPER for protocol based care of AIDS
- Page 20.717
- A common model for all the clinical protocol knowledge, specifies the concepts necessary to
define clinical protocols in a given domain of medicine
EON shows
1) How decision support systems can be embedded within larger clinical information systems
2) Exemplifies the use of emerging standards for network based communications among
software modules
3) The use of special purpose tools for entering and maintaining protocol knowledge base
- Page 20.731
- Formal Precedents for dealing with CDSS are lacking
- Whether the courts will use the systems under negligence law or product liability law
- Negligence law: a product or activity must meet reasonable expectation for safety
- Strict liability: a product must not be harmful
- Potential liability by MD who chose not to access CDSS, and made an incorrect decision when
the system would have suggested the correct care
- Page 20.731
- The Evaluation of CDSS is challenging
- Page 20.732
- Future Directions for CDSS
- WWW will bring CDSS into Patient’s home
- WWW will assist CDSS and how it affects both patients and MD
- Internet will affect the underlying technology with which CDSS are developed
- CDSS can be assembled from previously created, tested, and debugged components
- Page 20.733
- Mix and Match components for different purposes
Chapter 21: Computers in Medical Education - Page 21.737
Goals of Medical Education
1) Provide students specific facts and information
2) Teach strategies for applying this knowledge appropriately to the situation that arises in
medical practice
3) Encourage development of skills necessary to acquire new knowledge
- Page 21.738
- The Application of computer technology to education is referred to as computer assisted
learning, computer-based education (CBE), or computer based Instruction (CBI)
Advantages of using computers in medical education
1) To augment, enhance, or replace traditional teaching strategies to provide new method of
learning
2) An extension of student’s memory
3) Multimedia capabilities
4) Immersive interfaces: 3D, Force feedbacks
5) Support personalized one on one education
6) Anytime, any place, any pace learning
7) Exercise the student’s knowledge in a nonthreatening environment
8) More enjoyable and engaging, maintaining interest
- Page 21.739
- Historical look
- 1960’s in Ohio State U, Mass. General Hospital (MGH) and Univ of Illinois
- In OSU, the development of Tutorial Evaluation System (TES)
- At MGH, use CBE programs to simulate clinical encounters
- Page 21.740
- At Univ of Illinois developed Computer Aided Simulation of Clinical Encounter (CASE)
- Simulate clinical encounters between MD and patient
- The lack of portability of systems and the extreme expense of system development and testing
served as barriers to the widespread use of CBI
- Poor quality of transmissions and high costs limited access to CBE programs by distant users
- In 1983, the MGH programs were offered as the continuing medical education (CME)
components of the American Medical Associations Medical Information Network (AMA/NET)
- IN 1970: PLATO system at U of Illinois
- High cost of PLATO and the need for specialized terminals limit its use
- GUIDON: AI model of clinical reasoning
- Page 21.741
Computers Learning Methods:
1) Drills and Practice
2) Didactic: the Lecture
- Advantage: Removal of time and space
- Disadvantage: Cannot answer question
- Assess published reports immediately
3) Discrimination learning: the process that teaches the student to differentiate between the
different clinical manifestations
- E.g. dermatologic lesions
4) Exploration vs Structured Interface
- Exploratory environment in which students can experiment without guidance or interference
- Drill and Practice Advantages: Teach important facts without deviation
- Exploratory: Advantage: encourage experimentation and self discovery
- Disadvantage: Students may waste time
o Fail to learn important materials
- Page 21.745
5) Constrained vs Un-Constrained Response
- Disadvantages of Pre-defined set of response
a. Cues the student
b. Distracts from the realism of the simulation
- Advantages: Easier to work for developers
- Intermediate approach: Provide a single, comprehensive menu of possible actions, thus
constraining choices in a program specific, but not a situation specific manner
6) Construction: Learning anatomy through reconstructing the human body
7) Simulation
- Static: each case presents a patient who has a predefined problem and set of characteristics
- The underlying case remains static
- Dynamic: Simulate changes in patient state over time and in response to student’s therapeutic
decisions
- Page 21.746
- Immersive stimulated environment: Physical simulation of a patient in an authentic
environment such as an operating room
- Procedure trainers / Part Task Trainers
- Page 21.747
8) Feedback and Guidance
9) Intelligent Tutoring System: Mixed Initiative Systems
- Allow students freedom but provide a framework that constrains the interaction
- Coaching system: monitor the system and intervenes only when the student request help
- Tutoring systems: guide a session aggressively by asking questions
- Page 21.747
Current Applications:
1) Preclinical Applications
- E.g. Brain Storming: An interactive atlas of neuro anatomy, with images of dissections and cross
sections, diagrams, and extensive support text
- E.g. Digital Anatomist: 3D models of brain and anatomic structures
- E.g. Visible Human male and female through MLM
- HeartLab
- Page 21.748
2) Clinical Teaching Applications
- Page 21.749
- Allow rare diseases to be presented and allow the learner to follow the course of illness over
any appropriate time period
- Use of indestructible patient
- Cases developed can be shared
3) Continuing Medical Education
- Distance
- Costs
- Different Audiences
- Development of a just in time approach to deliver knowledge rich and problem focuses
information during the course of care
- Page 21.750
4) Consumer Health Education
- HC provider to suggest high quality web sites that can be trusted to provide valid information
- Page 21.751
5) Distance learning
- E.g. MedScape
- Challenges:
o Technical problems
o Disconnection
o No High speed connection
o Quality of program
- Page 21.752
Design, Development, and Technology: Design of Computer based learning applications
- 4 levels:
1) Structured Content:
- Early tools: both text and media, was embedded inside the program, along with the code that
presented content
- Pros: Contents developed by content experts, no need to know programming language
- Cons: Courseware difficult to maintain or modify
- Structured Content:
o Paragraph of a text: Narrative text
o Paragraphs broken into subsections, each represent a coherent concept, with the name
of concept as keyword, the paragraph is connected to structured text
o A database record, with fields, is a structured item
o Structured content adds labels or tags for semantic structuring
2) Query, retrieval, and indexing
- Students can access only the terms and links that are indexed
3) Authoring and Presentation
- An authoring system allows the expert to focus on the content of the teaching program
- Author is provided with a template
- Allow multiple authors to create content in parallel
- Content driven automated presentation systems are based on the use of structured content
- Page 21.755
4) Analysis and reasoning
- Built-in automated assessment of students
- Page 21.756
Application development
1) Definition of the need
2) Assessment of the resources
3) Prototyping and Formative Evaluation
- Focus groups with a facilitator can lead to useful design changes
- Formative Evaluation: Conducted during the evolution of a project at many stages
- Page 21.757
4) Production: the process of executing the design for the entire range of content
- A method of content collection and version control must setup
5) Integration in the curriculum
- Barriers:
a. Initial High cost of acquiring different resources
b. The reluctance of the faculty to modify their teaching
6) Maintenance and upgrade
7) Standards
- Page 21.758
- Standards for Meta data: the information that describes the content and add structure to the
content
- Content available to another groups
- Presentation and authoring programs developed in parallel
- Upgrade all to new
Technology Considerations
o Operating Systems
o Internet vs local machine
o Internet: Access to changing content
o Local machines: Need for high performance
- Page 21.759
Evaluation
- 4 levels:
1) Acceptability: Reaction of the student to the method
2) Usability
- 2 methods:
o Use of videotaped encounter with the program
o Use of automatically generated log of the student interaction
o Lac of information about the student’s motives for interaction
3) Knowledge Acquisition: Any impact on what the students learned
- Is the program more effective than traditional method of teaching the same material
- In what ways is computer based learning different from the traditional methods of learning?
4) Behavioral Change: Affect how students practice medicine
- Ultimate goal: Improve a student’s ability to solve problems through the application of that
knowledge
Chapter 24: The Future of computer applications in Biomedicine - Page 24.831
1) Low cost, high quality telemedicine
- Internet to support clear video images with high fidelity audio links
2) Remote consultation
- Improve access to expert patient care and enhance patient satisfaction
- Bring the experts to the patients via the Internet
3) Integrated health records
- Each record will be linked electronically over the internet so that each person has a single virtual
health record
- Record will be secure
4) Computer based learning
- Students use new immersive technologies
5) Patient and provider education
6) Disease management
- Page 24.833
Major challenges and opportunities facing the field of computational biology
1) A computational model of physiology
- But much of the pathophysiology needed are unknown
- Page 24.834
2) Design of new compounds for medical and industrial use
3) Engineering new biological pathways
4) Data mining for new knowledge
- Assumptions
- HC workers work more with computers
- Improvements in computer technology
- Page 24.835
- HC Cost
- Threat of malpractice continue
- Specialized Computer chips for medical applications
- Human machine interaction style changed
- Shift to optical networks
- Switch from analog to digital recording of info
- Electronic records will change the way we provide patient care and monitor health
- Page 24.836
- There will be an increased application of computers in all aspect of medicine
- Computers will become ubiquitous with high degree of interconnection and an increased ability
to operate
- Page 24.837
- Need an environment that brings together a large variety of computer based support tools
- Need the development of standards for data sharing and communication
- Better understanding of sociological issues
Computers integrated into medical settings by
1) Applications must fit the existing information flow in the settings where they are to be used
2) Computer systems should provide common access to all computer based resources , so a
user cannot tell where the program ends and another starts
- E.g. Radiologist using PACS to search for references
3) User interface must be both consistent across applications and easy to use
- E.g. Pointing, Natural Language interfaces
- Programs should have a common terminology to refer to frequently used concepts
- E.g. Diagnosis, symptoms, or lab test value
- Page 24.838
- Increasing amounts of medical information pooled into smaller, more powerful computers
- Future effects of nanotechnology
- Microelectromechanical systems (MEMS) build very small sensors and miniature mechanical
devices
- Page 24.839
Potential costs of using computers
o Cannot replace the flexibility of current paper based systems
o Progress notes about patients, using any words, in a any order, any format
- Page 24.840
- Computer based system
o Increased Legibility
o Increased accessibility
o Use of the information for other purposes such as clinical research
- Page 24.840
- Lawrence Weed noted that for medical records to be useful, they must be indexed so that
important information could be extracted
- Use Problem Oriented Medical Record (POMR)
- Medical records organized according to the patient’s current problem
- Standardization provides benefits but decreased flexibility
- Makes information more accessible but restrict freedom to pursue alternate means
- Page 24.841
- Computers remain decision support tools
- To monitor the quality of HC delivered
- Pharmacogenetics: Clinical and biological data combined to determine the individual response
to drug therapy
- Page 24.841
Factors affecting the role of computers in medicine
- Page 24.842
1) Changes in computers and biomedical technology
- Computers are smaller, less expensive, and more powerful
- Ability to connect multiple devices over high speed networks
- Multiple data storage devices are accessed by complex computers that sort and abstract patient
data
2) Changes in the background of health professionals and biologists
- Increase familiarity increases acceptance of computers
- Acceptance of the Internet model of interactions
3) Legal considerations
- Computers can provide reminders of overlook diagnosis
- Warnings that if ignore by HC workers, can bring lawsuits
- Page 24.844
4) Health care financing
- Slow the rate of growth of HC cost
- CDSS to assess which tests are most appropriate
- Used by Insurers to review MD’s decision
- Pharmacy Benefits Managers (PBM): Manage the prescription benefits for insured individuals,
working with pharmacy to reduce cost of care for insurers
-