2015 12 09 Minutes - CQF Data Model Call
Jump to navigation Jump to search
CQF Data Model Call Notes
- Wednesday December 9, 2015
- 1-2 PM ET
- +1 770-657-9270 Passcode: 217663
- URL: https://meetings.webex.com/collabs/#/meetings/detail?uuid=M8UL81KQZZKHCW46R28OYGINQM-8ENJ&rnd=47738.082690
- Bryn Rhodes
- Claude Nanjo
- Darrell Woelk
- Diedre Sacra
- Floyd Eisenberg
- Ken Kawamoto
- Nadia Ramey
- Rob McClure
- Tom Oniki
- Yan Heras
- Update to FHIR Quality (QI Core) with Logical View - Bryn submitted the Project Scope Statement to Dave Hamill and the FHIR Management Group to move forward with the planned update. There is no further information available at this time.
- Data Model
- Patrick Langford is working on an editor for CIMI archetype – the work is discussed on Friday meetings. For January, the goal is tooling that can read an ADL archetype and output a FHIR profile. It could make sense to create a single approach instead of the multiple efforts now in progress.
- Claude spoke with the FHIM group (Galen Mulrooney, Rob McClure) and discussed the possibility of focusing on the CIMI logical model for the FHIM. The mechanism needs to be discussed. The current effort is transformations from the FHIM model to FHIR. It may help to converge efforts to avoid duplicating work before the January meeting. Pending this discussion, the effort has paused a bit.
- Current Questions:
- What is the strategy for transformation from ADL to FHIR profiles? Is there a way to get from ADL to XSD? There is a way to get from ADL to XML. It should be possible to get to XSD. The purpose is that CQL tooling can use an arbitrary XSD and write CQL expressions from that (it is used to validate QUICK expressions). CQL is a good mapping environment. A map from the data model to XSD could allow mapping from FHIM to FHIR, etc.
- The Tuesday Q6 HL7 WG birds of a feather session in January could help collaborate to move forward. The Mapping is an expression translating instances from one model to another. The task is to import all data into tables from the source and write transformations to change the source data into the target. It involves intermediate mapping tables. CQL is more than a language to “get this” flexibly to define the transformations. It creates an “insert” statement so the data looks like the target – a series of insert statements. Each transformation returns a structure that looks like the target model. CQL creates a query to define an instance but doesn’t persist the instance directly. The issue is whether mapping tables are needed (might be handled using a terminology service). This represents (a) mapping of models Vs. (b) concept mapping. The two need to be stored in different ways and they are used in different ways. Concept mapping occurs at code system level (e.g., SNOMED-CT to ICD-9). Context mapping aligns complex objects. Context independent and context dependent mapping occurs through the terminology service.
- Plan to pick 5-6 model mappings and build a prototype to do it (not an instance of a model).
- Consider the invitation to display the existing work on the Tuesday Q6 birds of a feather meeting at the Orlando HL7 meeting. Bryn will continue work and inform Graham he will be presenting the prototype during the birds of a feather meeting in Orlando.
- Should we consider CIMI being adopted by the FHIM as the logical model and join with the FHIM efforts. Work then could move forward more collaboratively. Will discuss FHIM collaboration on a subsequent call when more FHIM members can join.
- Bryn will continue work on the prototype and reserve a place in the January WG Tues Q6 meeting birds of a feather session.
- The next meeting is Wednesday, December 16, 2015 at 1 PM ET. Communication will occur via email until that time.