Transcript PPT - Department of Computer Science

Jane Hunter
Hafsa Qureshi
 Jane





Hunter
M.Sc. eHealth candidate (part-time)
Undergrad in Computer Engineering at McMaster
Have also done an MBA at McMaster
Working as a software developer at NCR where
we are currently creating a Java development
infrastructure for next generation banking
products
Married for 10 years as of June, no kids
 Hafsa




Qureshi
M.Sc. eHealth candidate (full-time, thesis
student)
B.Sc. Health Informatics from University of
Waterloo
Trying to look for internship employment
Recently married
Introduce SNOMED CT
Show some of the uses of
SNOMED
Discuss some disadvantages
of SNOMED
Vocabularies
What is SNOMED CT?
Examples
Database Tables
Implementing an Application with SNOMED
Questions
SNOMED CT
• Systematized Nomenclature of Medicine Clinical
Terms
MeSH
• Medical Subject Headings
UMLS
• Unified Medical Language System
LOINC
ICD-10-CA -
• Logical Observation Identifiers Names and
Codes
• Enhanced Canadian version of the 10th revision
of the International Statistical Classification of
Diseases and Related Health Problems
 Systematized
Nomenclature of Medicine
Clinical Terms
 controlled medical vocabulary licensed
and supported by the International Health
Terminology SDO
 provides a common language that enables
a consistent way of indexing, storing,
retrieving and aggregating clinical data
across specialties and sites of care
 comprehensive, multi-lingual clinical
terminology that provides clinical content
and expressivity for clinical
documentation and reporting
 Medical
Subject Headings
 Comprehensive controlled vocabulary for
the purpose of indexing journal articles
and books in the life sciences
 Created and updated by the United States
National Library of Medicine (NLM)
 serves as a thesaurus that facilitates
searching
 used by the MEDLINE/PubMed article
database and by NLM's catalog of book
holdings
 Unified
Medical Language System
 compendium of many controlled vocabularies
in the biomedical sciences
 provides a mapping structure among these
vocabularies and allows one to translate
among the various terminology systems
 Comprehensive thesaurus and ontology of
biomedical concepts
 provides facilities for natural language
processing
 intended to be used mainly by developers of
systems in medical informatics
 Logical
Observation Identifiers Names and
Codes
 universal standard for identifying medical
laboratory observations
 developed and is maintained by the
Regenstrief Institute, a US non-profit
medical research organization
 publicly available at no cost.
 include medical and laboratory code
names, nursing diagnosis, nursing
interventions, outcomes classification,
and patient care data set.
 International
Statistical Classification of
Diseases and Related Health Problems
 provides codes to classify diseases and a
wide variety of signs, symptoms, abnormal
findings, complaints, social circumstances,
and external causes of injury or disease
 every health condition can be assigned to a
unique category and given a code, up to six
characters long
 Why
are there so many vocabularies?
Methods:
assembled 1929 source concept
records from a variety of clinical
information
records were coded in each scheme
by an investigator and checked by
the coding scheme owner.
codings were then scored by an
independent panel of clinicians for
acceptability

Conclusion:




No major terminology source can lay claim to being the
ideal resource for a computer-based patient record.
SNOMED is considerably more complete, has a
compositional nature and a richer taxonomy. It suffers
from less clarity, resulting from a lack of syntax and
evolutionary changes in its coding scheme.
READ has greater clarity and better mapping to
administrative schemes, is rapidly changing and is less
complete.
UMLS is a rich lexical resource, with mappings to many
source vocabularies. It provides definitions for many of
its terms. However, due to the varying granularities and
purposes of its source schemes, it has limitations for
representation of clinical concepts within a computerbased patient record.
 SNOMED





CT is
a clinical healthcare terminology
a resource with comprehensive, scientificallyvalidated content
essential for electronic health records
a terminology that can cross-map to other
international standards
already used in more than fifty countries
Each year, avoidable deaths and injuries occur
because of poor communication between
healthcare practitioners.
 The delivery of a standard clinical language for
use across the world's health information
systems can therefore be a significant step
towards improving the quality and safety of
healthcare.
 SNOMED CT aims to improve patient care through
the development of systems to accurately record
health care encounters.
 Ultimately, patients will benefit from the use of
SNOMED CT, for building and facilitating
communication and interoperability in electronic
health data exchange.

Consists of
over a million
medical
Concepts
• For example
22298006 means
myocardial
infarction (MI).
The Concepts
are arranged
in a type or
IS-A hierarchy
• For example,
Viral pneumonia
IS-A Infectious
pneumonia IS-A
Pneumonia IS-A
Lung disease.
 concepts
are organized in hierarchies, from
the general to the specific
 allows
very detailed (“granular”) clinical
data to be recorded and later accessed or
aggregated at a more general level
 Concepts
may have multiple parents, for
example Infectious pneumonia is also an
Infectious disease.
 The
Concept graph must be acyclic — a
parent cannot be its own child.
 From
abscess to zygote, SNOMED CT includes
more than 311,000 unique concepts.
 There
are almost 800,000 descriptions in
SNOMED CT, including synonyms that can be
used to refer to a concept.

a “concept” is a clinical meaning identified by a
unique numeric identifier (ConceptID) that never
changes.


formally defined in terms of their relationships with
other concepts.
give explicit meaning which a computer can process and
query on.
Every concept also has a set of terms that name
the concept in a human-readable way.
 ConceptIDs do not contain hierarchical or
implicit meaning.


The numeric identifier does not reveal any information
about the nature of the concept.
 Concept
descriptions are the terms or names
assigned to a SNOMED CT concept
 “Term” in this context means a phrase used
to name a concept
 A unique DescriptionID identifies a
description
 Multiple descriptions might be associated
with a concept identified by its ConceptID
 Some
of the descriptions associated with
ConceptID 22298006:





Fully Specified Name: Myocardial infarction
(disorder) DescriptionID 751689013
Preferred term: Myocardial infarction
DescriptionID 37436014
Synonym: Cardiac infarction DescriptionID
37442013
Synonym: Heart attack DescriptionID 37443015
Synonym: Infarction of heart DescriptionID
37441018
 Each
of the above descriptions has a
unique DescriptionID
Concepts are represented by a unique humanreadable Fully Specified Name (FSN).
 Each concept has one unique FSN intended to
provide an unambiguous way to name a concept.


not necessarily the most commonly used for that
concept
Each FSN ends with a “semantic tag” in
parentheses at the end of the concept to
indicate the semantic category to which the
concept belongs
 For example,



Hematoma (morphologic abnormality) is a FSN that
represents the description of what the pathologist sees
at the tissue level
Hematoma (disorder) is a FSN which indicates the
concept that would be used to code the clinical
diagnosis of a hematoma by a general practitioner.


Each concept has one Preferred Term meant to capture the
common word or phrase used by clinicians to name that
concept.
For example,




the concept 54987000 Repair of common bile duct (procedure)
has the Preferred term “Choledochoplasty” to represent a
common name clinicians use to describe the procedure.
Unlike FSNs, Preferred Terms are not necessarily unique
Occasionally, the Preferred Term for one concept may also
be a Synonym or the Preferred Term for a different
concept.
For example
Cold sensation quality (qualifier value) has a preferred term of
“Cold.”
 Common cold (disorder) also has a synonym of “Cold.”
 In both cases, “cold” represents a common clinical phrase used
to capture the meaning of the FSN.

 Synonyms
represent any additional terms
that represent the same concept as the
FSN.
 are not required to be unique across
concepts.
 Example:
 Some of the Synonyms associated with
ConceptID 22298006 which has the Fully
Specified Name: Myocardial infarction
(disorder) are:



Synonym: Cardiac infarction DescriptionID:
37442013
Synonym: Heart attack DescriptionID: 37443015
Synonym: Infarction of heart DescriptionID:
37441018
there are
about
1,360,000
links or

four types of relationships


Defining characteristics are IS_A relationships and
defining attributes.
Qualifying characteristics are non-defining, qualifying
attributes.


Historical relationships relate inactive concepts to active
concepts.


constrains the possible values an implementer can select in
assigning a qualifying characteristic to a concept
For example, a concept may be inactivated because it is a
duplicate. A “same-as” relationship would be created
between the 2 concepts.
Additional relationships are other non-defining
characteristics

For example, PART OF which is retained for backward
compatibility with SNOMED RT.
Each concept in
SNOMED CT is
logically defined
through its
relationships to
other concepts.
Every active
SNOMED CT
concept (except
the “SNOMED CT
Concept” Root
concept) has at
least one IS_A
relationship to a
supertype
concept.
establish IS_A
relationships with
one or more
defining concepts
(called
supertypes) and
modeling the
difference with
those supertypes
through defining
attributes.
 “Supertype-Subtype
relationships” or “Parent- Child
relationships.”
 IS_A relationships are the basis of the SNOMED CT’s
hierarchies.
Attributes relate two concepts and establish the
type of relationship between them.
 Example:


Lumbar discitis (disorder) (a concept in the Clinical
finding hierarchy) is related to concepts in the Body
structure hierarchy through two attributes: FINDING SITE
and ASSOCIATED MORPHOLOGY.




Lumbar discitis (disorder)
FINDING SITE Structure of lumbar intervertebral disc (body
structure)
ASSOCIATED MORPHOLOGY Inflammation (morphologic
abnormality)
The two attributes FINDING SITE and ASSOCIATED
MORPHOLOGY and their assigned values provide
definition for the concept Lumbar discitis (disorder).

Example:

the concept Pneumonia (disorder) is characterized with the attribute
FINDING SITE. Since pneumonia is a disorder of the lung, FINDING SITE
has the value Lung structure (body structure).


Pneumonia (disorder)
FINDING SITE Lung structure (body structure)
Clinical finding/disorder
Procedure/intervention
Observable entity
Body structurez
Organism
Substance
Pharmaceutical/biologic
product
Specimen
Special Concept
Physical Object
Physical force
Event
Environment or
geographical location
Social context
Staging and Scales
Concepts in this hierarchy represent the result of
a clinical observation, assessment or judgment,
and include both normal and abnormal clinical
states
 contains the sub-hierarchy of Disease.
 Examples of Clinical finding concepts:





Clear sputum (finding)
Normal breath sounds (finding)
Poor posture (finding)
Examples of Disease concepts:


Tuberculosis (disorder)
Non-Hodgkin's lymphoma (disorder)
 The
Event hierarchy includes concepts that
represent occurrences (excluding procedures
and interventions)
 Examples



of Event concepts:
Flood (event)
Bioterrorisk attach (event)
Earthquake (event)
 This
hierarchy contains such subhierarchies as Assessment scales and
Tumor staging
 Examples of Assessment scales concepts:


Glasgow coma scale (assessment scale)
Stanford Binet intelligence scale (assessment
scale)
 Examples


of Tumor staging concepts:
International Federation of Gynecology and
Obstetrics (FIGO) staging
Dukes staging system (tumor staging)
 provides
explicit links (cross maps)
to health-related classifications and
coding schemes in use around the world

e.g. diagnosis classifications such as ICD-9-CM,
ICD-O3, and ICD-10, as well as the OPCS-4
classification of interventions.
 Additional
cross-maps are also under
development or consideration
 Cross-maps facilitate reuse of SNOMED
CT-encoded data for other purposes, such
as reimbursement or statistical reporting
SNOMED CT is a multinational, multilingual
terminology
 has a built-in framework to manage different
languages and dialects
 The International Release includes a set of
language-independent concepts and relationships




available in US English, UK English, Spanish and Danish
Currently translations into French, Swedish, Lithuanian,
and several other languages
planning to translate the standard into other languages

This 85 year (258707000 year) old (70753007 old) (397659008
age) female (248152002 female) was admitted via the emergency
room (50849002 emergency room admission) from the nursing
home (42665001 nursing home) with shortness of breath
(267036007 dyspnea), confusion (225440008 onset of confusion),
and congestion (418092006 respiratory tract congestion). There
was no history of (14732006 no history of) fever (386661006
fever) or cough (49727002 cough) noted. Patient also has a
history of (392521001 history of) senile dementia (15662003
senile dementia) and COPD (13645005 chronic obstructive lung
disease).

Prior to (288556008 before) admission (129273005 admission –
action), the patient was taking the following medications:

Prednisone (116602009 prednisone), Lasix (81609008
furosemide), Haldol (349874003 oral haloperidol), and Colace
(418528006 docusate). Patient has also been taking Lorazepam
0.5-mg tablet (349865000 oral form lorazepam) 2x a day
(229799001 twice a day) as needed for anxiety (48694002
anxiety). Patient is also noted to have a vitamin C deficiency
(76169001 ascorbic acid deficiency).
 Is
SNOMED too complicated?
SNOMED CT is distributed as a set of tab-delimited
text files that can be imported into a relational
database.
the Concepts table, the Descriptions table, and the
Relationships table, are commonly referred to as
the “core” tables.
The association of a set of Descriptions and a set of
Relationships to each Concept is implemented using
the ConceptID which is the primary or foreign key in
the three tables
 The
Concepts Table contains all the concepts in
SNOMED CT
Concept ID
• Primary Key
Fully
Specified
Name
• Foreign key to description table;
serves to provide a human
readable name for each concept
Concept
Status
• Indicates whether a concept is in
active use or retired
 This
table relates the various terms used to
name a single SNOMED CT concept.
Descritption ID
• Primary key
Description Type • Fully specified name
(Indicates type • Preferred Term
of description) • Synonym
Language Code
• Associates each description with a
particular language or dialect

This table contains the relationships between SNOMED
CT concepts.
Relationship • Primary Key
ID
Relationship • Type of relationship
Type
• First concept in the
Concept ID 1 relationship
• “target” concept in the
Concept ID 2 relationship
The content of SNOMED CT evolves with each
release.
 Drivers of these changes include changes in
understanding of health and disease processes;
introduction of new drugs, investigations,
therapies and procedures; and new threats to
health, as well as proposals and work provided
by SNOMED partners and licensees.
 Changes designated as minor require only a
history record to record the change.
 The history mechanism involves the following
tables:



Component History Table
Component History References Table
A
Subset refers to a set of Concepts,
Descriptions, or Relationships that are
appropriate to a particular language, dialect,
country, specialty, organization, user or
context.
 The Subset Mechanism may be used to derive
tables that contain only part of SNOMED CT
 Subsets are not necessarily mutually
exclusive
 Subset mechanism involves the following
tables:


Subsets Table
Subset Members Table
 Cross
Mappings enable SNOMED CT to
effectively reference other terminologies
and classifications.
 Each cross map matches SNOMED
concepts with another coding scheme
that is called the “target scheme.”
 Cross Mapping mechanism involves the
following tables:



Cross Map Sets Table
Cross Maps Table
Cross Map Targets Table
 The
International Health Terminology
Standards Development Organization
provides technical documentation for
developing SNOMED compliant
applications
 Technical Reference Guide – 166 pages
 Technical Implementation Guide – 211
pages

Includes an appendix for working with HL7
(version 3)
All Version 3 products derive their semantic
content from the RIM.
 Intermediary models are used to constrain the
RIM for use in a particular specification.





Domain Message Information Model (D-MIM), which
constrains the portion of the RIM used by a committee in
the derivation of all their messages.
Refined Information Model (R-MIM), and a Hierarchical
Message Definition (HMD) specify a set of Message
Types.
there will be several D-MIMs derived from the RIM; there
will be several R-MIMs derived from each D-MIM; and
there will be several HMDs derived from each R-MIM.
SNOMED CT will typically apply these constructs
to these intermediary models, rather than to RIM
itself.
 The
RIM includes a new set of data types
developed for use within the HL7 Version
3 family of standards.
 Data types that can carry ConceptIDs
include:


Coded with Equivalents (CE), which carries a
code, the name of the coding scheme the code
is drawn from, and a display name
corresponding to the code; and allows
synonyms to be transmitted – such as an HL7
code and its equivalent SNOMED code;
Concept Descriptor (CD), which builds on the
CE by supporting the post-coordination of
codes (or, stated in another way, the combining
of codes from a terminology to create a new
concept).
RIM attributes of type CE or CD can also have a
specified vocabulary domain.
 These domains can include HL7-defined concepts
or can be drawn from HL7-recognized coding
systems such as LOINC or SNOMED CT.
 Vocabulary domains have a coding strength that
can be



“Coded, No Extensions” (CNE), in which case the only
allowable values for the field are those in the vocabulary
domain;
“Coded, With Extensions” (CWE), in which case values
other than those in the vocabulary domain (such as local
codes) can be used if necessary.
The vocabulary domain specifications stated in
the RIM always refer to a complete vocabulary
domain.
 at the RIM level there is no specialization based
on realm of use or on the context and needs of a
specific message.
 As RIM attributes are specialized to suit a
specific message context, the domain of the
attribute can be reduced (constrained) to reflect
the specialization.
 A vocabulary domain that has been constrained
to a particular realm and coding scheme (such as
SNOMED CT) is called a “value set.”

 Retrieve


an HL7 V3 message
A receiver of one or more HL7 messages will need
to be able to extract the coded information in
the message and aggregate it with concepts sent
in other messages or with concepts stored in a
data repository.
The entire discussion of aggregation in this
section assumes that valid and conformant HL7
messages are received.
 Syntactic


transformation
Convert the concepts in the message into a
“canonical form” (using a derived equivalence
between HL7 RIM attributes and SNOMED CT
relationship types).
The use of guidelines and templates can
constrain the inherent flexibility of an HL7
message, and can decrease the number of substeps required to perform a canonical
transformation. Tightly coupled systems can
take this into account, and establish bilateral
agreements that will minimize message
variability.
 Aggregation


Aggregate the various representations, all
expressed in a common canonical form.
Use techniques that query for the primary code
AND the semantic properties of the concepts of
interest.
 What
are the implications to system
designers when they try and hide the use of
SNOMED from the end user?







http://www2.infowayinforoute.ca/Documents/R2_ENGLISH%20SC%20Guide%20and%20S
tandards%20Catalogue.pdf
http://en.wikipedia.org (Medical_Subject_Headings,
International_Statistical_Classification_of_Diseases_and_Related_
Health_Problems, Unified_Medical_Language_System, LOINC,
SNOMED_CT)
http://www.ihtsdo.org/snomed-ct
http://eagl.unige.ch/SNOCat
Campbell JR, Carpenter P, Sneiderman C, Cohn S. Chute CG,
Warren J. Phase II Evaluation of Clinical Coding Schemes:
Completeness, Taxonomy, Mapping, Definitions, and Clarity.
Journal of the American Medical Informatics Association.
1997;4:238-51.
http://www.pubmedcentral.nih.gov/picrender.fcgi?artid=61239&
blobtype=pdf
SNOMED Clinical Terms User Guide