Stream: cql
Topic: CQl to ELM with FHIR
Jerry Goodnough (Dec 09 2019 at 16:34):
In working with CQL-ELM translator and FHIR what level of model validation is done? More broadly speaking is it possible to constrain the extensions to a specific set of profiles? I am guessing that at this point the answer is no. I am thinking that to get this I will need to walk the AST and analyze those elements that are extensions and identify what extension they actually are. Validation of more of a profile would likely be even more complicated.
Bryn Rhodes (Dec 09 2019 at 18:31):
The translator validates based on whatever is in the model info. The FHIR model info currently doesn't include any profile information. I am working on a version of the model info that does consider profiles. In that version, any extensions defined in a profile will be exposed as first class elements.
Bryn Rhodes (Dec 09 2019 at 18:33):
It still wouldn't validate extensions that are accessed directly (i.e. with .extension('extension-url').value), but the intent is that measure and decision support authors wouldn't need to use that pattern to access extensions, they will be accessed directly through first class elements for any extensions defined in the profiles.
Bryn Rhodes (Dec 09 2019 at 18:33):
So I wouldn't suggest validating profiles in the translator in that way, I've always considered it something that would be surfaced in the model info and validated that way.
Jerry Goodnough (Dec 09 2019 at 19:24):
So just curious it appears that the modelinfo file used is that same format for qdm and FHIR, how is the FHIR version generated.
Bryn Rhodes (Dec 10 2019 at 02:09):
The current versions are generated by xsd-to-modelinfo from the XSDs published as part of the FHIR specification.
Bryn Rhodes (Dec 10 2019 at 02:11):
The next versions will be generated by this tooling based on the actual StructureDefinitions published in the FHIR specification.
Last updated: Apr 12 2022 at 19:14 UTC
