– DRAFT –
FHIR RDF

Issue 94: Lack of concept URIs for CodeableConcepts (Leader: Gaurav)

gaurav: I've been looking at the UTG repo -- Unified Terminology Governance

https://github.com/HL7/UTG

gaurav: It used to generate terminology.hl7.org .
… That's the first place people should check. It also contains downloadable npm files, so this seems to be the nexus for these files. I'm trying to figure out which is the source of truth.

https://terminology.hl7.org/

gaurav: The npms are published in registry.fhir.org

gaurav: HL7 terminology package: https://registry.fhir.org/package/hl7.terminology|3.0.0

gaurav: It's referenced from https://build.fhir.org/terminologies.html#system
… It also explains when to have something added.
… The examples often show SNOMED, which is body parts based; and LOINC, which is procedures based.

eric: How does this compare with looking in UMLS? I guess this one is used for terminology supremacy.

gaurav: If you have codes in an existing code system, this allows you to bring it into FHIR.

eric: Canonical system identifiers for deployed terminologis?

gaurav: Yes.
… You can start with your own code and later map it to LOINC or something else.

Introductions

Introduction: Dagmar, working on data project with Inrupt, and read article from Eric about using JSON-LD in the next version of FHIR. Also part time PhD student on personal knowledge graph. Trying to use FHIR data and combine with other data.
… PhD student in Prague.

Issue 94: Lack of concept URIs for CodeableConcepts (Leader: Gaurav)

https://build.fhir.org/terminologies.html#system

http://tx.fhir.org/

https://github.com/HL7/UTG/issues/7

gaurav: Once I find that source of truth, I want to show a simple example. Then discuss it on zulip, where we'll probably get a lot of pushback.

gaurav: Google doc for making the case: https://docs.google.com/document/d/1sW3Tgj2J_wBzlUWih07e0Vf_M9Ue8YyPqjK0arhajS8/edit#

Semantics of Structure Definitions

eric: Observation has no semantics, right?

(Following up on last week's discussion https://www.w3.org/2021/12/16-hcls-minutes.html#t05 )

ben: When Observation has components, there's no value field at the top level -- they'll be blank at the top level.

Personal knowledge graph

(This brief discussion was in response to Dagmar's comment about being interested in the personal healthcare knowledge graph.)

gaurav: There’s a bunch of different RDF/Turtle files generated from the FHIR examples (see “RDF” at http://build.fhir.org/downloads.html#7.1 and https://www.hl7.org/fhir/downloads.html). I wonder if it would be useful to build a comprehensive knowledge graph out of those?

eric: I've wanted to do that for a while.

ADJOURNED