Scribenic: Claude Nanjo
There are a number of data models that map to the RIM. Some are conceptual and some are logical. ONC has been interested in two models in particular for CDS and CQI - the QDM and the vMR. Other models include CCDA, FHIR, DCM, etc... There has been a push to harmonize these initiatives. For CDS, an effort is under way to harmonize QDM and vMR into a single model. The vMR used to be more highly nested than it is today and thus a bit more V3-like. FHIR, on the
other hand, is a rather flat model and more intuitive to use for physicians. vMR lie in-between in terms of domain clarity. The RIM, on the other hand is far more abstract and more difficult to grasp.
The ONC selected the vMR for the Health eDecision initiative - Use Case 1 & 2. Use Case 1 consists of interoperable knowledge artifacts such as ECA Rules, Order Sets and Documentation Templates. Use Case 2 consists of CDS as a service.
Project Tacoma is the next step and looks at harmonizing the models and expression languages for HQMF and HeD Knowledge artifacts.
Another area of interest is how to incorporate much of the work done by Galen Mulrooney with FHIM. We have discussed this possibility as a group.
Another area of interest is to start moving towards V4 and adopt FHIR as the leaf model but add the necessary hierarchy needed for CDS.
We have considered a 'hybrid' approach whereby a CDS ontology could be added over the FHIR leaf model.
Yet the FHIR group has considered adding some degree of hierarchical structure to FHIR as well. The FHIR group has been very positive about the adoption of FHIR by CDS.
<eafry> CDS/CQI has 2-3 options for harmonization - must provide coverage - community doesn't what another model...only wants one
<eafry> option 1 - merge vmr and qdm
<eafry> option 2 - TACOMA option - harmonize at the conceptual model
<eafry> option 3 - develop conceptual model for fhir - collapse vmd and qdm into fhir resources
<eafry> option 3 generated most interest - FHIR willing to add hierarchy to current resources using "relationship" column to indicate parent/child relationships...both at the class level as well as at the property level
<eafry> question to use UML for this modeling effort ....or OWL
In a joint meeting with CDS and CQI, there was broad agreement about moving towards FHIR as a Logical Model. On the following day, in a meeting with the HL7 ARB, there were some concerns about using FHIR as the logical model as opposed to FHIR. There was also a concern in that FHIR does not formally define a UML model and the ARB wanted a UML model for the harmonized model. The community discussed the possibility of building either a UML model or OWL
starting with a few concepts to understand the implications of each choice. There may be some resistance at HL7 about using OWL as it is new but this resistance may be changing given the involvement of some FHIR members with semantic technologies.
Cognitive Medical System also led ballot reconciliation for three Service Functional Models (SFM) - Event Publish & Subscribe, Unified Communication, and Order Service. These three efforts also require a data model and ideally FHIR may be the right model, though new resources would need to be proposed.
Emory and Claude will reach out to Eric and Josh to see how to coordinate efforts in the definition of this prototype such that the produced artifact is complementary to their work.
Need a governance process around the FHIR extension process. Extensions may be for pure point-to-point, institutional-level, community-level, or national level. For, say, non-core national extensions, the ONC could play a role in the curation of these resources. FHIR will provide a framework for the extension process but it is the purview of some organization to ensure that extensions in their domain are consistent with the proper guidelines, have the right
structure, and correct semantics.