This presentation demonstrates the functionality provided by the Logical Model Designer (LMD) and Snow Owl tools, which enables terminology to be bound to the Singapore Logical Information Model.
Abstract:
A critical enabler in the journey towards semantic interoperability in Singapore is the Singapore "˜Logical Information Model' (LIM). The LIM is a model of the healthcare information shared within Singapore, and is defined as a set of reusable "˜archetypes' for each clinical concept (e.g. Problem/Diagnosis, Pharmacy Order). These archetypes are then constrained and composed into "˜templates' to support specific use cases.
The Singapore LIM harmonises the semantics of the information structures with the terminology, using multiple types of terminology bindings, including semantic, value domain and constraint bindings. Value domain bindings are defined to both national "˜reference terminology' (used for querying nationally-collated data), as well as to a variety of "˜interface terminologies' used within local clinical systems (required to enforce conformance-compliance rules over message specifications generated from the LIM). To support the diversity of pre-coordination captured in local interface terms, "˜design patterns' are included in the LIM, based on the SNOMED CT concept model. These design patterns represent a logical model of meaning for a specific concept, and allow more than one split between the information model and the terminology model to be represented in a semantically-consistent manner.
This presentation will demonstrate the "˜Logical Model Designer' (LMD) - an Eclipse-based tool that is being used to maintain Singapore's Logical Information Model. A number of features of the LMD tooling will be demonstrated, with a specific focus on how the information structure is bound to the terminology via an interface to the Snow Owl platform. Value Domains are defined as reference sets within Snow Owl and then linked to the information structures defined in the LMD.
Please see our website http://b2i.sg for further information.
What Goes Wrong with Language Definitions and How to Improve the Situation
The Logical Model Designer - Binding Information Models to Terminology
1. The Logical Model Designer (LMD) – Binding Information Models to Terminology
Linda Bird, Hendry Wijaya, Teck Wei Chin - MOH Holdings
Balazs Banfai – B2i
25th October 2012
(10:45 – 11:30)
2. Agenda
•Background
•Logical Information Model
•LMD Tooling
•Terminology Binding
•Demonstration
•Platform Software Architecture
•Project Timeline
3. Healthcare landscape of the future
Strategic vision of patients moving seamlessly across the
healthcare system, receiving coordinated patient-centric care
at the most appropriate settings.
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
RH
CH
Polyclinics NH
FPs Home
Care
Rehab &
support
services
Screening &
Prevention
Palliative
Care
Enabled by the National
Electronic Health Record
(EHR)
4. Sharing Information Across Singapore
Event Summary, Referral & Screening Documents
EHR Summary Record
Name
NRIC No.
Primary Care Provider
Care Coordinator
Allergies
Diagnoses
Current
Medications Procedures
Recent Events
Recent
Referrals
Care Plan
Date of Birth
Gender
Detailed Reports /
Images
Shared Services / Functionality
Business Intelligence
Data
Warehouse
National
Scorecard
Link
Share
Data
DA / ADR
Module
Immunisation
Medications
Reconciliation
Problem List
Reconciliation
Laboratory
Reports
Procedure
Reports
Radiology
Reports / Images
Drill Down Drill Down
SOC to AIC Referral
Community Hosp
Disch. Summ.
Referral to SOC
SOC
Event Summary
Inpatient
Disch. Summ.
GP
Event Summary
ED
Event Summary
Investigations
Immunisations
Applications that are not a part of the EHR solution,
but will be integrated with the EHR
Care Co-ordinator
Emergency Department
General Practitioner
Emergency Department
Acute-Care Hospital
General Practioner
Community Hospital
5. Use Case for Standards
oMessaging – safe exchange of transactional data
oDocument Exchange – safe exchange of documents
oPersistence – storing data in clinical systems
oInteroperability – ability to interpret semantics of data received from other clinical systems and store in native data stores
oQuerying & Analytics – over multiple heterogeneous sources
oDecision Support – ability to define and apply decision support rules over shared data
7. 1.A common, implementation-independent model of all shared healthcare information in Singapore.
2.A ‘single source of truth’ that articulates our clinician and business information requirements
3.Implementation artefacts are generated from common semantics
Logical Information Model
10. Archetypes
oReusable clinical models
oDefined as a set of constraints on the reference model
oFor example:
–Observation, Diagnosis, Alert, Adverse Reaction, Investigation Result
oCan be further constrained – e.g. Heart rate, Microbiology Result
Templates
oGroup together a set of archetypes for a specific use case
oApply additional use-case specific constraints
oCan be used to standardise the elements, constraints and terminology for a given message type, user interface etc
oFor example:
–Discharge summary, Investigation Report, Prescription
Archetypes and Templates
10
15. oCommon model
–shared healthcare information in Singapore
oCommon queries
–Provides a common vocabulary for querying multiple data sources
oCommon meaning
–Each clinical meaning is represented just once and reused many times
oConsistent terminology
–National reference terminology bound to a clinical meaning
oImplementation independent
–Information model independent of exchange format and proprietary information models
oMachine processable model
–Allows information to be defined once, and used many times (consistently)
oConformance Testing
–Enables the automation of Conformance Testing to ensure quality of data
Benefits of the LIM
15
16. Logical Model Designer
Clinician
Verification
Logical Reference Model (LRM)
ISO-13606 + ISO-21090
T
e
r
m
i
n
o
l
o
g
y
S
e
r
v
e
r
Clinical Archetypes
Logical-Object Model (LOM)
Constrained Clinical Archetypes
Logical-Object Model (LOM)
Template Definition
Logical-Object Model (LOM)
Compiled Template Hierarchy
Logical-Object Model (LOM)
Implementation Artefact
Logical-XML, HL7 v2, Java
Documentation Artefact
Word, Excel, HTML
Snow
Logical Model Designer (LMD) Owl
Value sets
Examples
Constraints
Meaning
17. Logical Model Designer
Clinician
Verification
Logical Reference Model (LRM)
Composition, Entry, Cluster, Element
T
e
r
m
i
n
o
l
o
g
y
S
e
r
v
e
r
Clinical Archetypes
Observation
Constrained Clinical Archetypes
Heart Rate
Template Definition
Discharge Summary
Compiled Template Hierarchy
Discharge Summary Hierarchy
Implementation Artefact
Discharge Summary LXML + Java CC Testing
Documentation Artefact
Discharge Summary Hierarchy (Excel, HTML)
Snow
Logical Model Designer (LMD) Owl
Value sets
Examples
Constraints
Meaning
18. oValue set binding
•National: For national queries over heterogeneous data
•Local: For generating conformance/compliance software for messages populated using local value sets (mapped to national)
oConstraint binding
•Assumed, fixed and default values
•OCL and ESCG constraints
oSemantic binding
•Enables querying data using more general or specific meanings
•Enables searching models, and identifying semantic overlap
oRelationship bindings
•Defines meaning of relationship between parent to child node
•Used to construct design pattern bindings (aka constructor bindings)
oDesign pattern bindings (aka ‘constructor bindings’)
•Extended compositional grammar expression with path-based slots
•Used to normalise data that may either be precoordinated or postcoordinated in the structure, into a canonical form
Types of Terminology Binding
19. Terminology Binding Syntax
oUniform Resource Identifier (URI) standard:
•Simple and extensible means for identifying a resource (ftp:, mailto:, news:, etc.), in this case terminology artefacts
•Logical identifier of a resource, without specifying the physical location and the access mechanism
•Both human and machine readable, with the syntax: <scheme name>:<hierarchical part>[?<query>][#<fragment>]
•Obtaining a representation of the resource is done by Resolvers
oScheme for terminology binding: terminology:<code system id>[:version]?<query type>= <query expression>[&<extension key>=extensionvalue]*
oSNOMED CT query types:
•concept, conceptlist, refset, refsetlist, escg, ocl
21. Binding Terminology to a Coded Datatype
Concept Id
2.16.840.1.113883.6.96
SNOMED CT
20110701
Preferred Term
Description Id
Assumed value: PT
RefSet Id
RefSetVersion
SNOMED CT
22. Binding Terminology to a Coded Datatype
Code
2.16.840.1.113883.6.3
ICD-10
2008
Description
Assumed value: Desc
Value Domain Id
Value Domain Ver
ICD-10
23. Binding Terminology to a Coded Datatype
Code
DD OID
MOH Data Dictionary
20120101
Description
Assumed value: Desc
Value Domain Id
Value Domain Ver
DD Value Domain
24. oSnow Owl
̶Reference sets and Extension concepts
oLogical Reference Model (LRM)
̶Editable view
oArchetype – Observation ENTRY
̶Add semantic and value domain bindings
̶Node meaning, relationship meaning, constraint & constructor bindings
oConstrained archetype – Heart rate ENTRY
̶Add semantic and value domain bindings
̶Confirm subsumption validation
oCompile template – Discharge summary (with heart rate)
̶Fill slot with heart rate archetype, and
oModel search
̶Search models using semantic bindings
Demonstration of LMD Tooling
25. oBuilt on the seasoned Eclipse tooling platform with wide industry adoption
–Composed of bundles running within an OSGi (Eclipse) container
–Bundles can be deployed depending the product definition (possible for both client and server side)
–Help and branding information are in separate bundles
oUtilizes the services provided by the Snow Owl terminology platform
oCore domain objects are modeled via standard EMOF as opposed to be hard-coded
Platform Software Architecture