Stream: genomics
Topic: coding grouper resource
Patrick Werner (May 05 2020 at 09:19):
I just tried to use the grouper resource for capturing "relevant/favorite variants" = Variants the tumor board team decides on that these are important.
Grouper doesn't have a code to qualify it, i think it should have one. My workaround for now would be List Resource with a code "relevant Variants".
Patrick Werner (May 05 2020 at 09:19):
Proposal: add a qualifying component to the grouper to be able to express what kind of grouper it is.
Kevin Power (May 05 2020 at 12:13):
Seems like a reasonable request - log a JIRA?
Bret H (May 05 2020 at 12:44):
I thought OandO had discussed other options? was grouper to determined to be the best option?
Patrick Werner (May 05 2020 at 12:47):
Didn't saw that discussion. Had a discussion about notes.
Kevin Power (May 05 2020 at 12:48):
@Bret H - I found this? https://chat.fhir.org/#narrow/stream/179256-Orders-and.20Observation.20WG/topic/DiagnosticReport.20Sections
Kevin Power (May 05 2020 at 12:51):
And this one? https://chat.fhir.org/#narrow/stream/179256-Orders-and.20Observation.20WG/topic/DiagnosticReport.20scope
Bret H (May 05 2020 at 13:02):
thinking more like From: "clingenomics@lists.hl7.org" <clingenomics@lists.hl7.org>
Date: Monday, March 18, 2019 at 9:33 AM
To: "clingenomics@lists.hl7.org" <clingenomics@lists.hl7.org>
Subject: Re: [clingenomics] Block Vote for Tuesday, March 19
Hi all,
Sorry for this confusion here, but I need to pull a tracker from the block to discuss tomorrow (Tuesday, March 19).
19937 Panel+is+allowed+but+it+is+confusing (Clement McDonald)
Proposal from FHIR subgroup: Persuasive with mod --
-
Add textual guidance pointing out the two approaches to nesting information (nested/referenced reports AND nested/grouped observations)
-
Change name on panel to show it is NOT currently 1-1 with ordering panels (“grouper”)
-
Change cardinality of recommended-action reference (should be 0..*)
-
File a separate tracker for how to align best with O&O on the ordering side (to confirm grouper could be used with serviceRequest to align with the common notion of “panel” where labs want to report out specific panels)
Reason why? We actually voted at the WGM to remove the “Panel” observation 😊. It was one of the few we officially voted on.
https://gforge.hl7.org/gf/project/fhir/tracker/?action=TrackerItemEdit&tracker_item_id=16108&start=0
Summary
Genetics Panel: Suggest removing
Resolution
WGM Jan 2019
Persuasive with Mod - will remove, but need guidance on how to use diagnostic-report summary-of extension
Bret H (May 05 2020 at 13:04):
Kevin Power (May 05 2020 at 13:08):
Looks like after the migration to JIRA, it is now J#16108, and we switched it to "Not Persuasive"
Patrick Werner (May 05 2020 at 13:09):
we had many discussions around this. IIRC we came to the conclusion that panel was missleading as it implies a possible panel order. But we still saw the need to group observations. So we renamed it to grouper
Bret H (May 05 2020 at 13:12):
Ok. Looked through and did not see anything other than talking about labeling sections in diagnostic report., regards coding a type in grouper observations.
Bret H (May 05 2020 at 13:14):
To me the question of how to communicate grouping was still open. IF grouper is assumed to be the best choice, then a way to provide some indication of what the section, err I mean, grouper is about can be useful.
Bret H (May 05 2020 at 13:15):
However, list of variants the tumor board wishes to see....How about 'highlighted variants'? Why wouldn't the tumor board conclusion come back as a diagnostic report with a section containing the 'important' variants?
Patrick Werner (May 05 2020 at 13:16):
report would not solve many (my) use-case.
In my usecase (tumo board) you look at all OBS of a DR. Then choose your "favourites" which are the basis of your recommendation. Also this is used by the person preparing a meeting. To capture all "relevant" Variants for this case.
Jamie Jones (May 05 2020 at 13:17):
We voted to allow "both" uses of grouper - arbitrary collecting and grouping to represent a physical panel. As I recall we still don't want membership in a grouper to convey additional meaning to the variants within it
Patrick Werner (May 05 2020 at 13:18):
Bret H said:
However, list of variants the tumor board wishes to see....How about 'highlighted variants'? Why wouldn't the tumor board conclusion come back as a diagnostic report with a section containing the 'important' variants?
highlihted sounds nice, thanks.
The end result of a board is a DR, but during working on a case you don't have a DR
Patrick Werner (May 05 2020 at 13:18):
Jamie Jones said:
We voted to allow "both" uses of grouper - arbitrary collecting and grouping to represent a physical panel. As I recall we still don't want membership in a grouper to convey additional meaning to the variants within it
good summary.
Jamie Jones (May 05 2020 at 13:19):
You could use grouper.note to convey a note about the grouping
Patrick Werner (May 05 2020 at 13:19):
Would say "no somantical meaning which can be expressed otherwise"
Jamie Jones (May 05 2020 at 13:19):
Like "Patrick's favorite group"
Patrick Werner (May 05 2020 at 13:20):
I'd prefer a coding please.
Patrick Werner (May 05 2020 at 13:20):
Much easier to query
Kevin Power (May 05 2020 at 13:20):
Note versus a coding with no binding isn't a lot different :slight_smile: (mostly kidding)
Jamie Jones (May 05 2020 at 13:21):
Main issue is we don't want folks to use groupers instead of populating fields on individual obs or inserting needed references to other resources
Patrick Werner (May 05 2020 at 13:21):
in the downstream IG it will be a huge difference
Patrick Werner (May 05 2020 at 13:22):
Agree Jamie. Thats what i meant with "no somantical meaning which can be expressed otherwise".
Jamie Jones (May 05 2020 at 13:22):
Could you profile Grouper and require a different code?
Patrick Werner (May 05 2020 at 13:23):
nope
Bret H (May 05 2020 at 13:23):
@Patrick Werner your use case is part of workflow? Patrick said "The end result of a board is a DR, but during working on a case you don't have a DR"
Jamie Jones (May 05 2020 at 13:23):
*Additional code should be possible but messy
Patrick Werner (May 05 2020 at 13:23):
Observation.code is 1..1 would be fine with multiple codings here, but not possible
Bret H (May 05 2020 at 13:24):
seems like the use of grouper proposed here is midstream in a process (which would make it a problem internal to a reporting system) and not in the final report
Patrick Werner (May 05 2020 at 13:24):
yes
Jamie Jones (May 05 2020 at 13:25):
In that case we force folks to use an extension or drive their own profile mirroring grouper...
Kevin Power (May 05 2020 at 13:25):
I was wondering if this is an extension versus component?
Bret H (May 05 2020 at 13:25):
@Patrick Werner your use case is part of workflow? Patrick said "The end result of a board is a DR, but during working on a case you don't have a DR" seems like the use of grouper proposed here is midstream in a process (which would make it a problem internal to a reporting system) and not in the final report
Jamie Jones (May 05 2020 at 13:27):
We don't have great discipline in deciding which concepts become components/values vs extensions... Generally we just put everything in components because they are simpler to implement but I do wonder
Bret H (May 05 2020 at 13:29):
again. I am trying to understand Patrick's use case. He appears to be talking about PRE-report internal processes. why would that be the purview of anyone other than the system carrying out the process? is the use case clearly valid?
Kevin Power (May 05 2020 at 13:31):
OK - maybe I missed in the discussion, sorry. Is this 'group category' important / relevant to other receiving systems?
Jamie Jones (May 05 2020 at 13:32):
I worry that adding a cc to grouper to denote why the observations are grouped in it opens up room for shoehorning in semantic meaning
Patrick Werner (May 05 2020 at 13:35):
You are right Bret, this is mostly an internal Problem. But internal for us are also decentralized TumorBoards, with different participants sharing the data across several systems.
Jamie Jones (May 05 2020 at 13:35):
If we do our homework and determine what use cases are really improved by explicitly stating the grouping intent we may be fine to add it in
Patrick Werner (May 05 2020 at 13:36):
Just wanted to bring this to the group as i thought this could be also interesting to others. Right now we can group, but can't express why we grouped them.
Patrick Werner (May 05 2020 at 13:36):
I'm fine to have this only internal in our system
Jamie Jones (May 05 2020 at 13:37):
It sounds like prime extension territory to me, but other groups may want to use it as well
Kevin Power (May 05 2020 at 14:04):
I think a local extension for this makes sense for now, and we can see if others bring similar use cases forward.
Last updated: Apr 12 2022 at 19:14 UTC
