Meeting minutes
introductions
Spencer: Embedded systems, model driven eng. First job was research at Vanderbilt, interop approach to sys integration and mil avionics. In defense for several years. In other roles worked on tech standards for data modeling for integration for airplances.
… Recently moved into healthcare. Worked at Center for Med Interop, on med devices.
… Now soft eng for a data science team in healthcare strategy space.
… Trying to finish PhD.
Data Model Drive System Integration (Spencer Crosswy)
spencer: Was involved in FACE (Futre Airborne Capability Env) project. Software integ model was unsustainable.
… Led to vendor lockin. Led to FACE initiative.
… Key part was a data model based approach.
… Defined a data modeling language, similar to FHIR structure defs.
… Defines E-R model. Delivered as machine-readable spec of the data.
… Hugely beneficial. But still requires a human to understand model differences.
… FHIR feels very similar.
… Huge feature is to adapt to local env, w extensions and profiles.
… Interested in the diffs of FHIR profiles and how to deal with them.
… Want to map a solution to FACE and FHIR.
… Interested in ideas and collaborators.
http://
david: Also suggest thinking about "Model integrity" and "Suitability for use" as described in that paper.
david: Suggest looking at that as a way to think about the interop problem.
david: Might also look at ontology mapping space.
gaurav: One useful place to start with ontology mappings might be the SSSOM format, but I’m not sure how relevant it will be to your problem: https://
Issue 94: Lack of concept URIs for CodeableConcepts (Leader: Gaurav)
https://
Example here: https://
gaurav: In some cases the system and code can be concatenated to form a URI, but not in all cases.
… One option is to use a service to map a system/code pair to a URI.
… Or we could define a standard mapping. But SNOMED has some cases that are more complex to map.
david: How may coding systems do we need to handle?
gaurav: IDK. Could look in the examples.
david: There can always be new ones added.
david: I llike the idea of defining standard mappings, perhaps in the FHIR profile repo.
david: I don't like the service idea, because it means doing a request to it, and it means maintaining it.
Action: Gaurav to explore the use of the FHIR registry for standard system/code mappings to URIs
james: Some hospitals have their own coding systems too.
FHIR References
spencer: One thing I've noticed missing is the type of references.
https://
Created this as issue 95: https://
Action: Spencer to look at issue 95
ADJOURNED