Stream: implementers
Topic: new StructureDefinition based on DomainResource with Forge
Georg Fette (Sep 17 2019 at 12:13):
When creating a new StructureDefinition with Forge (v. 22.0), why isn't it possible to base the new profile on the base profile DomainResource ? I would like to create a custom profile that is completely independent of the exiting profiles. Or is DomainResource so abstract, that my idea does not make sense at all ? (btw. is there a more specific stream for Forge related questions than "implementers" ?)
Ward Weistra (Sep 17 2019 at 12:50):
Hi Georg,
I think you're likely right that this is because DomainResource (and Resource) are abstract base resources, but I'll check with @Michel Rutten. Can you share the use case you're looking to create a profile for?
You're also welcome to ask Forge questions in the #tooling channel. We're considering a dedicated stream for the Firely toolsuite, I'll put your vote on that :)
Best,
Ward
Lloyd McKenzie (Sep 17 2019 at 13:00):
What's your usecase for profiling DomainResource
Michel Rutten (Sep 17 2019 at 13:22):
Hi @Georg Fette, if you derive from Resource or DomainResource, you are defining a custom resource. Currently, Forge does not support custom resources.
Lloyd McKenzie (Sep 17 2019 at 19:37):
Not necessarily. You could theoretically profile DomainResource to set expectations around the use of meta, extension, etc. and "instantiate" that profile in other resources (e.g. Observation, AllergyIntolerance). Is that the use-case @Georg Fette ?
Ewout Kramer (Sep 17 2019 at 20:14):
@Michel Rutten You're definining a new resource when derivation=specialization, but it would be a re-useable constraint (which is what Lloyd refers to) when derivation=constraint. I read Georg's text ("create a custom profile that is completely independent") as wanting to create a new custom resource.
Georg Fette (Sep 18 2019 at 06:56):
My usecase is that I want to algorithmically transform a bunch of openehr archetypes and templates into fhir data model definitions. I do not need to specifically derive from anything, but as it seemed convenient to give the new structuredefinitions a basedefinition, I thought that domainresource could be nice. The resulting custom profiles are not compatible to anything from the rest of the fhir cosmos but at least I can use all tools and servers that are on the fhir side to handle my data, as from a specification point of view it will be fhir.
Michel Rutten (Sep 18 2019 at 11:57):
@Lloyd McKenzie indeed, that is possible but also not currently supported by Forge.
Lloyd McKenzie (Sep 18 2019 at 12:38):
@Michel Rutten It sounds like that's not what Georg needs so we'll leave it as a theoretical possible future requirement.
@Georg Fette If you're defining logical models, you can't start those with DomainResource (because they're not FHIR resources). If you wanted to define a logical model that looked similar to DomainResource and had similar elements and inherit from that, you could
Frank Oemig (Sep 20 2019 at 13:06):
within v2+ this also a consideration/use case. Why not creating logical models that have the same characteristics like resources, although they are not "official"?
Next we would like to generate an API for it...
Michel Rutten (Sep 20 2019 at 13:18):
@Frank Oemig You cannot use Logical models to exchange data. You can only exchange instances that conform to an actual resource definition. You can implement an interface based on custom resource definitions. Of course such an interface would not be FHIR compliant.
Frank Oemig (Sep 20 2019 at 13:31):
you cannot instantiate a logical model with data?
Grahame Grieve (Sep 20 2019 at 13:44):
there's actually 2 answers to this: you cannot instantiate a logical model with data and be FHIR conformant. But you can instantiate a logical model with data and be conformant to something else, whatever the logical model is expressing
Grahame Grieve (Sep 20 2019 at 13:44):
e.g. CDA or V2
Last updated: Apr 12 2022 at 19:14 UTC
