Stream: terminology
Topic: https in CodeSystem URI
Patrick Werner (Dec 16 2019 at 17:01):
The CG WG currently wants to establish an system uri for the NCI Thesaurus. This was already done by the mcode project, they decided to use: https://ncit.nci.nih.gov
As the CG IG an mcode trying to align to each other we now would adapt this URI.
The question is: Is it ok to have https in the CS URI? https://www.hl7.org/fhir/terminologies-systems.html currently only shows one CS having a https URI.
So our decision is using a (unusual) URI with https and mcode could stay like it is vs. change it to http
Grahame Grieve (Dec 16 2019 at 19:31):
it's not wrong, but I prefer not to
Michael Lawley (Dec 16 2019 at 22:45):
The OWL file indicates http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl as its URI if I'm not mistaken
Michael Lawley (Dec 17 2019 at 20:48):
We have it loaded in our sandbox server (see https://r4.ontoserver.csiro.au/fhir/CodeSystem?_summary=true&url=http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl ) as an example @Patrick Werner but I'm aware that there's also an NCI Thesaurus OBO edition and am not sure how we should treat that; is it the same code system but just a different version, or is it a different code system altogether? i.e., should we treat lexically equal codes from the two terminologies as equivalent/equal?
Grahame Grieve (Dec 17 2019 at 20:51):
hmm. have both you read this?
Grahame Grieve (Dec 17 2019 at 20:51):
http://hl7.org/fhir/ncimeta.html
Grahame Grieve (Dec 17 2019 at 20:51):
presumably mCode had not either....
Michael Lawley (Dec 17 2019 at 20:56):
Bur that is for the NCT Metathesaurus not the NCT Thesaurus
Grahame Grieve (Dec 17 2019 at 21:03):
oh. what's the difference?
Michael Lawley (Dec 17 2019 at 21:13):
The website (https://ncimeta.nci.nih.gov/) is not specially helpful, but the text below reveals that the Metathesaurus is an aggregation of other things, similar to UMLS, and not a code system itself.
"NCIm contains most public domain terminologies from the National Library of Medicine's UMLS Metathesaurus, as well as many other biomedical terminologies created by or of interest to NCI and its partners"
NCI Thesaurus, on the other hand does look like a code system:
"NCI Thesaurus (NCIt) provides reference terminology for many NCI and other systems. It covers vocabulary for clinical care, translational and basic research, and public information and administrative activities."
Grahame Grieve (Dec 17 2019 at 21:31):
it's probably more important that we describe the use of the NCI Thesaurus then?
Michael Lawley (Dec 17 2019 at 21:34):
I would think so. NCIt is a standalone thing. But guidance on NCIm would also be useful. As would guidance on use of any OWL-based terminologies since there's often a URI already selected.
Michael Lawley (Dec 17 2019 at 21:35):
But, once we start with OWL, we'll be drawn in to OWL terminologies that import other OWL terminologies and the lack of support for this by the CodeSystem resource.
Grahame Grieve (Dec 17 2019 at 21:36):
I don't recall what the lack of support for this is?
Michael Lawley (Dec 17 2019 at 21:37):
Because parent and child properties are code-typed, it is not possible to have hierarchy that spans code systems
Grahame Grieve (Dec 17 2019 at 21:37):
that's both true and false.
Grahame Grieve (Dec 17 2019 at 21:38):
hang on - no, we said we were going to fix that.
Michael Lawley (Dec 17 2019 at 21:39):
This paper discusses the issues we faced: https://knowledge.amia.org/69862-amia-1.4570936/t004-1.4574923/t004-1.4574924/3202087-1.4575024/3202087-1.4575025?qr=1
Grahame Grieve (Dec 17 2019 at 21:40):
AMIA members only
Rob Hausam (Dec 17 2019 at 21:41):
I'm an AMIA member and I didn't get anything when I followed the link
Michael Lawley (Dec 17 2019 at 21:43):
It should give you a login page, but then you need to re-follow the link after logging in
Rob Hausam (Dec 17 2019 at 21:44):
it didn't in Chrome, but I'll try copy and paste in Firefox
Carmela Couderc (Jan 03 2020 at 14:04):
https://jira.hl7.org/browse/FHIR-12675? and https://jira.hl7.org/browse/FHIR-17778? (where @Keith Boone shares the OID for NCI Thesaurus. )
Jay Lyle (Jan 06 2020 at 22:00):
Not working for me. A title might help.
Michael Lawley (Jan 06 2020 at 22:32):
FHIR OWL: Transforming OWL ontologies into FHIR terminology resources
- Alejandro Metke Jimenez ; Michael Lawley ; David P. Hansen
Patrick Werner (Jan 08 2020 at 10:48):
The OWL file indicates http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl as its URI if I'm not mistaken
In GF#12675 you are proposing the usage of http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl which would be very unusual for a canonical CS URI. I would prefer: http://ncit.nci.nih.gov
In my understanding The OBO Edition of NCIT uses the same concepts and concept codes therefore i would use the same URI for both editions.
Michael Lawley (Jan 08 2020 at 14:16):
To me this issue is that we are not the owners of NCIt so we don't get to choose the URI for someone else's CodeSystem, and since there's already one chosen as evidenced by the OWL definition of the CodeSystem content, then that's the best choice.
I'm not across the details of the "OBO Edition", but if it is the "same" CodeSystem, then it should probably have the same URI (but a different version). To determine whether it is the "same", a good question to ask is "should we consider two lexically equal codes, one from each edition, to be the same"?
Robert McClure (Jan 08 2020 at 21:25):
I hate to keep punting these sorts of things to HTA, but that is the HL7 entity that needs to set the policy for something like this. I see two different issues here, one is the standard "What is the canonical url for this external code system?" but the other is a policy issue for a code system that has multiple representations. I am wondering if this already is an issue for SCT - is there an owl representation that has different constructs but the same content as the RF2 version? If multiple formalisms of the same code system exist and each changes with a new "version" of the code system, does that mean each code system release produces multiple versions?
Grahame Grieve (Jan 10 2020 at 03:20):
that is the HL7 entity that needs to set the policy for something like this
It should start with what we already have. And when is it going to do it?
Michael Lawley (Jan 30 2020 at 02:37):
To answer the SCT question, there is no conflict between OWL and RF2 (nor should there be; the URI is for the logical / conceptual code system, not for a specific rendering of it)
Patrick Werner (Mar 11 2020 at 15:01):
Michael Lawley said:
To me this issue is that we are not the owners of NCIt so we don't get to choose the URI for someone else's CodeSystem, and since there's already one chosen as evidenced by the OWL definition of the CodeSystem content, then that's the best choice.
Yes the owner of the resource should ideally decide the used canonical url, but nci hasn't decided on an uri for their CS.
I don't like the http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl as it points to a file, which is unusual für CS URIs. Also the Link doesn't resolve. I'm still in favour of http://ncit.nci.nih.gov
How long does the HTA process last? Was there any progress?
Last updated: Apr 12 2022 at 19:14 UTC
