Stead Presentation

8/13/2019 Stead Presentation

1/28


2/28

Burning platform: overwhelming complexity

Setsoffa

ctsperdecisio

n

1000

10

100

5Human cognitivecapacity


3/28

Burning platform: overwhelming complexity

Setsoffa

ctsperdecisio

n

1000

10

100

5Human cognitivecapacity

2000 20101990 2020

Structural genetics:

e.g. SNPs, haplotypes

Functional genetics:Gene expression

profiles

Proteomics and other

effector molecules

Decisions by clinical

phenotype

(Adapted from) Stead WW. Beyond expert-based practice. IOM (Institute of Medicine). Evidence-based

medicine and the changing nature of health care: 2007 IOM annual meeting summary, (Introduction andOverview, p. 19). Washington, DC: The National Academies Press 2008.


4/28

Todays healthcare non-system

Experts practice by

working around

systems

System

development


5/28

Science Evidence Care

Missed Opportunities, Waste, and Harm

Institute of Medicine. Best Care at Lower

Cost: The Path to Continuously Learning

Health Care in America. Washington, DC:

The National Academies Press, 2013.


6/28

one in which science and

informatics, patient-clinicianpartnerships, incentives, and

culture are aligned to

promote and enable

continuous and real-timeimprovement in both the

effectiveness and efficiency

of care

Institute of Medicine. Best Care at Lower

Cost: The Path to Continuously Learning

Health Care in America. Washington, DC:

The National Academies Press, 2013.

Drivers:

System approach to care

Harness system for

discovery

Link discovery back into

system

Learning healthcare system


7/28

System approach to healthcare

Compassion

Pattern

recognition

Judgment

People

Simplification

Standardization

Process+ Informatics+

Reproducible

performance

Systems=

Memory

dependence

Forcing

function


8/28

System

development

System-supportedpractice

Individualize & act Assess

Plan Order

Workflow Peoples roles

Process

Technology tools

Evidence Research

Guidelines

Practice database

Pick population Risk

Cost

Variability

System approach to healthcare

Stead WW. Beyond expert-based practice. IOM (Institute of Medicine). Evidence-basedmedicine and the changing nature of health care: 2007 IOM annual meeting summary, p. 96.

Washington, DC: The National Academies Press 2008.

Monitor & correctProcess Patient Sentinel events

Process outcomes

Clinical outcomes

Status

Results

Trends


9/28

Ventilator Associated Pneumonia (VAP)Data from all Vanderbilt adult ICUs combined


10/28

Using informatics to make process visibleVentilator management dashboard

10


11/28

Ventilator management bundleData from all Vanderbilt adult ICUs combined

Development

Just-in time Adherence Data Feedback

Design Studio

Educaton


12/28

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

5

10

15

20

25

30

35

Jan-06

Feb-06

Mar-06

Apr-06

May-06

Jun-06

Jul-06

Aug-06

Sep-06

Oct-06

Nov-06

Dec-06

Jan-07

Feb-07

Mar-07

Apr-07

May-07

Jun-07

Jul-07

Aug-07

Sep-07

Oct-07

Nov-07

Dec-07

Jan-08

Feb-08

Mar-08

Apr-08

May-08

Jun-08

Jul-08

Aug-08

Sep-08

Oct-08

Nov-08

Dec-08

Jan-09

Feb-09

Mar-09

Apr-09

May-09

Jun-09

Jul-09

Con

currentBundleC

ompliance(%)

VAP

Rate

per1000

VentDays

Monthly VAP Rate Overall VAP Rate Bundle Compliance

Education

Development

Just-in time Adherence Data Feedback

Design Studio

Ventilator Associated Pneumonia (VAP)Data from all Vanderbilt adult ICUs combined


13/28

VAP rates per 1000 ventilator days:All Vanderbilt adult ICUs combined

23.6

20.1

29.3

21.5

17.5 17.9

13.8

10.8

7.5

4.53.3

0

5

10

15

20

25

30

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Ra

te

per1000

Ve

ntDays

Dashboard Implementation:

8/07-12/07

VAP Definition Retired by CDC December 2012


14/28

Central conclusions

Current efforts aimed at nationwide deployment of HCITwill not be sufficient to achieve the vision of 21stcentury

health care, and may even set back the cause

Success will require emphasis on providing cognitive

support.

In the near term, embrace measureable health care

quality improvement as the driving rationale for HCITadoption efforts.

Principles to support change

Record all available data to drive care, process

improvement, and research

Architect information and workflow systems toaccommodate disruptive change

Archive data for subsequent re-interpretation

Seek and develop technologies that clarify the context of

data

1/2009

Healthcare IT expectation gap


15/28

data mining

automation

connectivity decision

support

Root cause: mismatch between

computational technique & scale of problem

Stead WW. Electronic Health Records. In: Rouse WB, Cortese DA, eds. Engineering the system of

healthcare delivery. Tennenbaum Institute Series on Enterprise Systems, Vol. 3. Amsterdam: IOSPress; 2010.


16/28


17/28

Example - Electronic Health Record as a platform for research

De-identification

Clinical

notes

Physicianorders

Patient and staff

messaging

Billing

codes

Labs, Radiology, test results

Syntheticderivative

Electronic Health Record

~ 2.2 million records

>1.4 with adequate clinical data

De-identified DNA repository152.95k adult samples

20k pediatric samples

>37.6k with dense genetic data

Discarded blood

samples from

routine testing

VanderbiltBioVU

If eligible,

extract DNA

De-identification
http://www.onlinetelemedicine.com/html/product/sam_images/X-Ray.jpg


18/28

Clinical Notes(NLP - natural language

processing)

Billing codes

ICD9 & CPT

Medications

ePrescribing

& NLPLabs & test results

NLP

Finding phenotypes in the EHR

True cases

Finding phenotypes in the EHR


19/28

Algorithm Development and Implementation

Identify

phenotype

of interest

Case & control

algorithm

development and

refinement

Manual

review; assess

precision

Deploy

with

BioVU

Genetic

discovery &

replication

95%


20/28

Example of EHR discovery

Step 1 - Find individuals with normalhearts

Normal ECG

time

No heart disease

No Na-blocking drugs

No abnormal K, Ca, Mg

Hypothetical Record

n = 5,272 across

5 eMERGE sites

Step 2 - Find genetic variants associated with QRS duration in normal hearts

QRS duration is a continuous variable

in the normal range

Adjusted for age, sex, BMI


21/28

GWAS of QRS DurationSCN5A/SCN10A

n=5,272

Ritchie et al. Circulation 2013;127:1377-1385


22/28

Step 3 - Find phenotypes associated with the genotype of interest

Pull EHRs of individualsWith SCN10A (rs6795970)

Extract clinical phenotypes

(~ 1,600 & controls)

Construct Phenome-wide association scan

(PheWAS)

Denny et al Bioinformatics 2010; 26:1205-1210

n = 13,617 across5 eMERGE sites


23/28


24/28

Step 4In silico trial of what happens in a population with normal hearts

Normal ECG

time

No heart diseaseNo Na-blocking drugs

No abnormal K, Ca, Mg

Hypothetical Record

Myocardial infarction Atrial fibrillation

Follow the n=5272 with initial

normal cardiac phenotype for

development of atrial

fibrillation based on genotype

Ritchie et al. Circulation 2013;127:1377-1385


25/28


26/28

Workstream 2: Computation & informatics research

Privacy enhancing

architectures & algorithms

Computation across diverse

data sources

Vision: adding non-biologic determinants of health

Feature extraction &

qualification algorithms

Contributions:

Hypotheses

Feature extraction algorithms

Phenotype signatures

Pattern detection &

prediction algorithms


27/28

Structured/ unstructured

personal, health, health

care & community data

sources

Features to collect or

extract

Determinant

signatures

Determinant specific

decision support

End game vision: executable knowledge to support whole

person population health & health care

Executable knowledgeDiscovery platform

Privacy enhancing,

feature extraction &

pattern detection

algorithms

Curated

data sets

Life & health & healthcare management

apps

Clinical trials &

routine use


28/28

OLD NEWOne integrated set of data Sets of data from multiple sources

Capture data in standardizedterminology

Capture raw signal and annotate withstandard terminology.

Single source of truthCurrent interpretation of multiple

related signals

Seamless transfer among systemsVisualization of the collective output of

relevant systemsClinician uses the computer to update

the record during the patient visit.

Clinician & patient work together with

shared records and information.

The system provides transaction-level

data.The system provides cognitive support.

Work processes are programmed andadapt through non-systematic work

around.

People, process and technology worktogether as a system.

Next generation EHR computational paradigm

Stead WW. Electronic Health Records. In: Rouse WB, Cortese DA, eds. Engineering the systemof healthcare delivery. Tennenbaum Institute Series on Enterprise Systems, Vol. 3. Amsterdam:

IOS Press; 2009.

Stead Presentation

Documents

Transcript of Stead Presentation