QIDAM
Issues and proposals for a logical model
For discussion during HL7 WG Meeting in Jan 2014
Thursday Q3
Background
Need to harmonize the “clinical data model” for
Clinical Decision Support
Clinical Quality Information
Currently, there are multiple specifications that address these domains
Virtual Medical Record
Quality Data Model
Health Quality Measures Framework (Data request in r2)
Clinical Statements and associated templates in HQMF, QRDA, CCDA
Quality Improvement DAM
Conceptual model that aims to harmonize the domains
Represented as a UML class diagram
Concepts modeled mostly by drawing from vMR, FHIR, QDM
QIDAM was submitted for ballot for the Jan 2014 cycle
The scope was incomplete – not all clinical concepts were modeled
Will add significantly to the scope for the May cycle
Next steps
How do we get to a logical model?
What is the foundation of the logical model?
V3
FHIR
VMR
Other?
Integration of the logical model into the existing specifications
Two aspects of CQI
SPECIFYING THE KNOWLEDGE AND
USING IT IN REASONING
APPLYING THE KNOWLEDGE TO A
PATIENT
Knowledge specifications and
interoperability
Patient data interoperability and exchange
Examples
Examples
Measure specs: HQMF
CDS artifacts: Knowledge Artifacts (Health
eDecisions), Arden Syntax MLMs
Data models
Used in expressions about patient data
QDM
VMR
Reasoning model
Quality reports: QRDA
Transitions of care: CDA
CDS: VMR, CDA
Data models
Used in instances of patient data
Clinical statement and CCDA/QRDA templates
VMR and templates
Data exchange model
Reasoning model versus data exchange
model
Reasoning model requirements/desiderata may be different than for data exchange
model
Important example: a reasoning model must be compact so that expressions in
these approaches can be easy to read, write, and implement
Many data exchange models already
V2
V3
QRDA, CCDA
VMR
FHIR
Expressions in reasoning model and in data exchange
model
In QDM:
In Clinical Statement pattern:
Diagnosis, Active
Act [classCode=“ACT” and moodCode=“EVN”]
using “Asthma Value Set”
code = (“LOINC Code for Problem”) and
sourceOf[typeCode=“COMP”]
observation[classCode=“OBS” and moodCode=“EVN”]
code=“SNOMED-CT Code for Problem”
value = “Asthma Value Set”
sourceOf[typeCode=“REFR”]
observation[classCode=“OBS” and moodCode=“EVN”]
code=“LOINC Code for Status”
value=“SNOMED-CT code for Active”
SAIF Model
QIDAM
QI Logical
Model
Clinical
Statement
VMR Logical
Model
VMR ITS, CCDA, QRDA,
HQMF
Proposal
Focus on a reasoning model
Where there is a need, created mappings between the relevant data exchange model
and the reasoning model
E.g., We must map from QRDA to reasoning model used in HQMF
Advantages
Addresses the immediate issue in harmonization of HQMF and HeD Knowledge
artifacts
Does not create yet another data exchange model
Issues in creating the reasoning model
What form should the reasoning model take?
1. Quality Improvement Virtual Record (QIVR)
UML model that is a refinement of QIDAM
Possibly replaces or refines the vMR
2. FHIR profile
Can this meet the needs of a reasoning model?
3. A RIM refinement
4. Others?
What is the role of templates?
Recommended approach
Create a FHIR profile from QIDAM
Create a hierarchical logical model equivalent to the FHIR profile
Used in expressions
Advantages
Reuse of models/profiles/resources created by other WGs
E.g., medication-related resource created by Pharmacy
Avoids creating another model
Tooling of FHIR
Mappings of FHIR to CDA family
RDF mapping – enables CDS