Fundamentals of snomed ct
-
Upload
sb-bhattacharyya -
Category
Healthcare
-
view
820 -
download
0
Transcript of Fundamentals of snomed ct
SNOMED CT FUNDAMENTALS
Dr SB Bhattacharyya
MBBS, MBA, FCGP
Member, National EHR Standardisation Committee, MoH&FW, GoI
Member, IMA Standing Committee for IT, IMA Headquarters
Member, Health Informatics Sectional Committee, MHD 17, BIS
President (2010 – 2011), IAMI
SNOMED CT
■ SNOMED CT is a semantically interoperable polyhierarchical subtype multi-lexical clinical terminology system that is able to deliver robust benefits to the health care services.
■ SNOMED CT is a reference terminology for clinical terms and consists of terms used in health and health care.
Dr SB Bhattacharyya© 2
Why SNOMED CT?
■ Natural language for end-users
– Doctors
– Dentists
– Nurses
– Paramedics
■ Since clinical documentation systems document the clinical thoughts of the observers, a concept-based system works best, not a term-based or classification-based one
■ For clinical analytics, evidence based medicine, clinical decision support, etc., the correct idea is vital, not a prefixed term
Dr SB Bhattacharyya© 3
SNOMED CT
In the full version of SNOMED CT 2014 July International Release there are:
Components Active Total
Concepts 300,751 403,836
Description (terms) 1,037,903 1,206,870
Relationships 908,668 2,412,104
Term : Concept Ratio 3.45+ –
Term : Relationships Ratio 3.02+ –
Dr SB Bhattacharyya© 4
Advantages
Through its structure and support for refset (reference sets) or subsets, extensions and maps it permits:
■ Local additions of concepts and terms that are not present in the International Release
■ Mapping to other health care codes like ICD, LOINC, local codes (aka interface terminology)
■ Clinical pathway automation
■ Running extremely rich clinical analytics even on free text unstructured data
Dr SB Bhattacharyya© 5
Advantages
■ Users get to choose that particular term or a set of terms that they are most comfortable with in their own language/dialect
■ Users can express their clinical thoughts very precisely using expressions
■ Many of the record management tasks can be automated leading to extremely user-friendly richly-ergonomic systems
■ Rule-based alerts and warnings can be triggered more accurately
Dr SB Bhattacharyya© 6
Disadvantages
■ Learning the clinical terminology systems requires patience and perseverance – although this is applicable only to system designers and architects
■ Composing and querying expressions is a technical challenge –this too is again only applicable to system designers and architects
Dr SB Bhattacharyya© 7
SNOMED CT Components – Core
■ Concepts
– Concept Ids :: meaningless machine-processable numbers that act as unique identifiers for each
■ Descriptions (Terms)
– Human-processable terms :: each concept has a uniquely human-readable Fully Specified Name (FSN) and one preferred term for each language/dialect apart from synonyms
■ Relationships
– Between concepts: source to destination :: provides the “intelligence” for machines to process
Dr SB Bhattacharyya© 8
SNOMED CT Concepts
■ SNOMED CT is a concept-based code system based on the principle “one code per concept and one concept per code”, where concept is a thought or idea
■ Every clinical term is “expressed” as a concept and represents a “unit of meaning”
■ When a user chooses a clinical term, he is choosing the particular term that best expresses his thought or idea regarding a particular item, which could be a complaint or an item of past history or family history or an observation related to physical examination or investigation or diagnosis or treatment, etc.
Dr SB Bhattacharyya© 9
SNOMED CT Terms
■ Fully Specified Name (FSN) – identified by having the term suffixed by its top-level hierarchy enclosed in parenthesis thus -“(disorder)”, “(body structure)”, “(procedure)”, etc.
■ Preferred Term (PT) – one for each language/dialect
■ Acceptable (Terms) – synonyms
Dr SB Bhattacharyya© 10
SNOMED CT Relationships
■ Used to link concepts to help define it
■ Other than the root concept, a single concept will have at least one subtype relationship link and none-to-many (0…1) attribute relationship links that provide a unique definition for the concept
■ Provides the “intelligence” that enables IT systems to “comprehend” the meaning of a concept so that the most appropriate action may be initiated using rule-based, machine learning algorithms, etc.
Dr SB Bhattacharyya© 11
SNOMED CT Logical Model
Dr SB Bhattacharyya© 12
SNOMED CT Relationships
Infective pneumonia (disorder)
Lung (body structure)
Infection (disorder)Respiratory disease (disorder)
Virus (organism)
Viral pneumonia (disorder)
Dr SB Bhattacharyya© 13
Concept Model
■ Shaped in an inverted tree fashion with root concept at top representing the coarsest concept to the finest leaf at the bottom representing the finest concept
■ Each concept has a corresponding definition that is either primitive or fully defined and expressed using
■ Each concept (other than the root concept,) must have at least one | is a |meronomic relationship with another concept that is its supertype parents or ancestors
■ Each concept may have none-to-many (0…n) attribute “has a” relationship with other concepts
■ Each attribute has a name-value pairing in the domain name = range valueformat
Dr SB Bhattacharyya© 14
SNOMED CT Concept Model (schematic)
Dr SB Bhattacharyya© 15
SNOMED CT – Attribute Value-Range
DOMAIN(hierarchy)
ATTRIBUTERANGE
(concept of hierarchy)
<< This concept or one of its
descendants| Clinical finding | | FINDING SITE |
<< 442083009 | Anatomical or acquired body
structure |
| Body structure | | LATERALITY | < 182353008 |Side|
< Descendants only. Not the concept itself
Dr SB Bhattacharyya© 16
SNOMED CT Expressions
■ Clinical thoughts are “expressed” using SNOMED CT
■ Actual “code” of SNOMED CT, ideally hidden from all users and handled exclusively by the system
■ Consists of either of the following two formats
– ConceptId
– ConceptId | Description (Term) |
■ Needs to adhere to compositional grammar rules and expression constraint syntax
Dr SB Bhattacharyya© 17
SNOMED CT Expressions
■ Pre-coordinated
– Coordination already done
– Representation of a clinical meaning using a single concept identifier is referred to as precoordination
■ Post-coordinated
– Coordination done on-the-fly by the users choosing various rules-based alternatives
– Representation of a clinical meaning using a combination of two or more concept identifiers is referred to as postcoordination
Dr SB Bhattacharyya© 18
Pre-coordinated Expression –Example Clinical thought: Viral pneumonia
Expressed as:
ConceptId Term
75570004 viral pneumonia
The advocated format can be any one of the following two:
1. 75570004
2. 75570004 | viral pneumonia |
Dr SB Bhattacharyya© 19
Post-coordinated Expression –Example Clinical thought: Family history of hypertension
No single pre-coordinated term exists for this.
The clinical thought needs to be expressed as a composite of the following individual pre-coordinated concepts:
ConceptId Term
281666001 family history of disorder
246090004 associated finding
64572001 disease
38341003 hypertension
Dr SB Bhattacharyya© 20
Post-coordinated Expression –Example [contd.]
A postcoordinated expression based on the concept 281666001 | family history of disorder | can be used to record a family history of any disorder.
The definition of this concept includes 246090004 | associated finding | = 64572001 | disease | and the value of this attribute can be refined to 38341003 | hypertension | (which is a subtype descendant of 64572001 | disease |).
Dr SB Bhattacharyya© 21
Post-coordinated Expression –Example [contd.]
Therefore, the following postcoordinated expression can be created as below and used to represent this family history –
281666001 | family history of disorder | : 246090004 | associated finding | = 38341003 | hypertension |
This can also be written as –281666001 : 246090004 = 38341003
Dr SB Bhattacharyya© 22
Greatest Impact Area
Clinical data analytics
■ All aspects of a medical record can now be analysed using computers
■ This leads to the discovery of many hidden facts that can be statistically correlated and
■ Using Machine Learning techniques predictive analytics can be used to support personalised medicine
Dr SB Bhattacharyya© 23
Introduction to SNOMED CT by Dr SB Bhattacharyya available at
■ www.amazon.com/Introduction-SNOMED-CT-S-B-Bhattacharyya/dp/9812878939/ref=sr_1_1?ie=UTF8&qid=1453269681&sr=8-1&keywords=introduction+to+snomed+ct – hardcopy only
■ www.amazon.in/Introduction-SNOMED-CT-2016-Bhattacharyya/dp/9812878939/ref=sr_1_1?ie=UTF8&qid=1453269722&sr=8-1&keywords=introduction+to+snomed+ct – hardcopy only
■ www.springer.com/gp/book/9789812878939 – ebook and hardcopy
■ link.springer.com/book/10.1007/978-981-287-895-3 – chapter-wise online access only
Dr SB Bhattacharyya© 24
Dr SB Bhattacharyya© 25
Dr SB Bhattacharyya© 26
Thanks!