FHIR Chat · Post-coordination · terminology

Stream: terminology

Topic: Post-coordination


view this post on Zulip Grahame Grieve (Sep 29 2016 at 00:50):

so I spent some time on the plane working on support for post-coordinated terms. and I came up with a question: does it make sense to allow or disallow the use of post-coordinated codes in ValueSet.compose.include.filter.value

view this post on Zulip Grahame Grieve (Sep 29 2016 at 00:50):

?

view this post on Zulip Michael Lawley (Sep 29 2016 at 01:29):

If the op relates to subsumption (isa / is-not-a / generalises), then yes it does make sense to allow it

view this post on Zulip Michael Lawley (Sep 29 2016 at 01:31):

But it does raise the question of whether the op '=' means "equivalent" (which would require subsumption testing) or "identical" (which would only require lexical equality).

view this post on Zulip Grahame Grieve (Sep 29 2016 at 02:35):

yes it raises several questions. but why do you think it makes sense to allow it?

view this post on Zulip Grahame Grieve (Sep 29 2016 at 02:36):

what's a coherent use case?

view this post on Zulip Michael Lawley (Sep 30 2016 at 00:03):

Any time there isn't an appropriate pre-coordinated code - eg finding : laterality = side

view this post on Zulip Grahame Grieve (Sep 30 2016 at 00:04):

but that would be more conveniently 2 filters:
concept is-a finding
laterality = code

view this post on Zulip Michael Lawley (Sep 30 2016 at 00:17):

Sorry, I keep losing my messages due to the esc key being right next to the backtick & I was hurried in my re-typing.
The correct post coordinated expression for this contains nesting and is
finding : finding_site = ( body_structure : laterality = side )

view this post on Zulip Grahame Grieve (Sep 30 2016 at 00:30):

yes but you could do all that with filters and it would be more convenient

view this post on Zulip Rob Hausam (Oct 03 2016 at 21:30):

You could do this example with one or more "simple" filters. That approach probably is going to meet most of the typical cases, but there are more complex cases where that wouldn't work. So I think we either need to decide to limit our support to the "simple" cases, or we'll need to allow the post-coordinated expressions (and support that approach).

view this post on Zulip Grahame Grieve (Oct 03 2016 at 21:32):

well, the filter that is relevant to me is expression = [constraint]. I think that it's a bad idea to support 2 different things with that sophistication; make it one or the other


Last updated: Apr 12 2022 at 19:14 UTC