Stream: implementers
Topic: Group
Grahame Grieve (Oct 24 2017 at 05:04):
Some questions about group:
- How do you search the server for all the patients who are in a particular group?
- there's no binding for Group.characteristic.code. Surely we should have a value set that at least includes all patient demographics fields, and observations, conditions, and allergies
- shouldn't value/exclude be a repeating group to allow and + or pattern (our usual pattern)
- why is there no binding on Group.code? Do we even need it?
- The definition of Group.actual doesn't really help me. I happen to know what Entity.determinerCode is, so I think I know what it does, but that's a rather esoteric fact to know these days
- is quantity a definitional or descriptional aspect of a group? It sounds like a descriptional thing to me, and something you'd ask the server about (see point 1) (at a particular point in time, in fact)
@Lloyd McKenzie @Brian Postlethwaite
Grahame Grieve (Oct 24 2017 at 09:00):
And what does it mean if a group has both members and characteristics?
Grahame Grieve (Oct 24 2017 at 09:50):
<Group xmlns="http://hl7.org/fhir">
<active value="true"/>
<type value="person"/>
<actual value="true"/>
<name value="Test Group"/>
<characteristic> <!-- 0..* Trait of group members -->
<code>
<coding>
<system value="http://hl7.org/fhir/StructureDefinition/Patient"/>
<code value="Patient.gender"/>
</coding>
</code>
<valueCodeableConcept>
<coding>
<system value="http://hl7.org/fhir/ValueSet/administrative-gender"/>
<code value="male"/>
</coding>
</valueCodeableConcept>
<exclude value="false"/>
</characteristic>
</Group>
Grahame Grieve (Oct 24 2017 at 09:51):
is that the right way to say 'all males'?
Grahame Grieve (Oct 24 2017 at 10:30):
that works on my server now
Lloyd McKenzie (Oct 24 2017 at 14:14):
1. Group?_id=123&_include=Group.member
2. We could come up with an example binding I suppose. Allergies wouldn't make much sense I don't think. We'd definitely want to add species as veterinary is a primary use-case
3. Don't understand. Current model lets you say "include males" or "exclude males". If you want to say "include males" + "include other", then you'd repeat characteristic twice.
4. If we can come up with a value set that covers all animals, materials, devices, etc., then I guess it could have a binding
5. This differentiates "Crate of chickens #5" vs. "All chickens within 5 miles of intersection XYZ"
6. It depends on the value of the actual flag. If you're defining a study group, you could say "50 male mice between 6-9 months old" with actual = false and have it be definitional. Or you could say "Test group A contained 48 male mice between 6-9 months old" with actual = true and have it be descriptional
7. It means that you've defined the group by characteristics but identified the matching candidates - e.g. you've now figured out which chickens fell within the 5 mile radius
Grahame Grieve (Oct 24 2017 at 19:52):
1. only includes explicitly listed members
2. policy is that we should
3. according to the spec, characteristics are anded not ored. there's no way to or conditions
5. umm, I don't see that as a meaningful difference for a group
6. I think that needs to be very clear. the documentation is not useful right now
7. we should definitely document that too
as it is, group is uninterpretable because of documentary short comings. In particular, we need to provide some documentation about how to use Group.characteristic.code
Grahame Grieve (Oct 24 2017 at 21:31):
I upgraded my server. It now supports a _group parameter on search that allows you to restrict matches to items in the group - either explicitly, or by characteristic. e.g.
GET http://test.fhir.org/r3/Patient?_group=102
Grahame Grieve (Oct 24 2017 at 21:33):
but for now, the only characteristic that works is the gender one from the example above
Christiaan Knaap (Nov 01 2017 at 19:09):
Thanks for trying this out Grahame.
From the spec it is not even clear to me whether the characteristic may be expressed a SearchParameter (as I read it in the example above), or whether other types of characteristics are possible. And for searchparameters I think a simple search url would be an easier representation (but still not capable of 'or').
Am I thinking too simple with a solution where
- characteristic consists of a search url and include / exclude,
- characteristic is 1-*, where the includes are unioned (thereby enabling 'or') and the excludes are excluded afterwards?
Grahame Grieve (Nov 01 2017 at 22:43):
actually, I moved away from expressing as a search parameter. but my server only supports elements that have an associated search parameter
Brian Postlethwaite (Nov 10 2017 at 06:47):
The one that was missed above was the count, that's intended to be a size of the group. Which may not necessarily be the actual count of members listed to do inactivity, or people moving out of the group.
So you can then cover things like the number of places in the AA group session, but the actual number of members could differ day to day.
(We would use it as an assist in scheduling group sessions)
Brian Postlethwaite (Nov 10 2017 at 06:47):
Then also when someone leaves, you know that there is a place that you can add to the group.
Brian Postlethwaite (Nov 10 2017 at 06:48):
(A kind of soft value)
Maybe some clarity on this would be useful.
At most it _could_ be an informative constraint.
Shaishav Bhatt (Dec 28 2020 at 10:38):
I need to convert my backend symptom data into FHIR resources, backend is in Python, how should I go about it? which libraries shall I use? I do not need to communicate with any servers or import any resources, what should I consider?
René Spronk (Dec 28 2020 at 10:45):
@Shaishav Bhatt Please create a new topic, you've now attached your question to an old 2017 discussion. If you'd like to use Python to create some sort of mapping, you may wish to ask the #python stream. The generic topic of mapping is probably best discussed on this forum, but please create a new topic for it. Thanks!
Last updated: Apr 12 2022 at 19:14 UTC
