Stream: genomics
Topic: Study Level
Kevin Power (Jan 03 2022 at 20:12):
Sorry I missed the meeting today, but I saw these notes from today's meeting:
“Study-level” Observation proposal https://jira.hl7.org/browse/FHIR-32084
FHIR Genomics Study Level Proposal (https://docs.google.com/document/d/1N9JsMW3OXVO1VVipIYs-Cz6J0zpWEx51yvigMGlINOo/edit )
Following up with O&O on discussion re: adding Procedure to the DiagnosticReport workflow. Similar to ImagingStudy
Question on best extension to reference from the report (0..*)
Changes to GenomicsBase (abstract profile)
Changes to RegionStudied
May not need range-based components if sending BED directly
Changes to OverallInterpretation
What was the group's thinking on using a profile of Procedure to represent the 'Performed Genetic Study'?
@Jamie Jones @Bob Dolin @Daniel Rutz @Bret H
Bob Dolin (Jan 03 2022 at 20:18):
Hi @Kevin Power , we're not sure if @John David Larkin Nolen has discussed this with OO, but we thought that would be a good next step.
Jamie Jones (Jan 03 2022 at 20:27):
Other than that it was sort of "stunned silence". I brought up that I felt the use of "supporting-info" was overloaded and it should have its own extension. Dan suggested the extension on DR could be optional since the link to Procedure is already present on the observation.partOf field
Bret H (Jan 03 2022 at 20:34):
my perspective was that it was the first time some had thought about it. important note: Clem as well as others thought Procedure was a good choice to profile off of. The major objection at this point is complexity. some discussion on possible ways to overcome the complexity were put forward. I don't think we had time (nor spent it) on pointing out the more academic, but pertinent, concepts of Definitional objects versus Instance events. The document has that discussion in it. So, my take, there will be more for folks to say if/when they read the document.
Kevin Power (Jan 03 2022 at 20:46):
+1 to the overload of supporting-info extension.
I will defer to the expertise of O&O, and I might be the only one, but this still feels weird. Does anyone else in the lab space use Procedure like this? It sort of feels like we are admitting that we really need a new resource, but because of the complexity of getting that approved, we are settling for profiling a resource that doesn't really fit our use case? It still strikes me that we are declaring we will not use nearly 1/2 of the attributes of Procedure in our profile (assuming that is what these strike-thru's mean?)
Bob Dolin (Jan 03 2022 at 20:54):
@Kevin Power We had similar discussions based on the RIM - a 'procedure' in the RIM more closely aligns to SNOMED's notion of a 'surgical procedure', whereas SNOMED's notion of a 'procedure' more closely align's to the RIM's 'ACT'. FHIR has broadened the use of Procedure, to include 'an activity that is performed on, with, or for a patient as part of the provision of care', with examples such as 'adult day care services', 'exercise', etc.
Bob Dolin (Jan 03 2022 at 20:55):
For me, I'm not so concerned about using Procedure, but more that it's unusual for a lab report to include metadata about the service itself. It's common for imaging, uncommon for lab, but genomics is pushing the envelope to where I think we need something.
Kevin Power (Jan 03 2022 at 20:56):
I think this part in the Procedure page is where I am most tripped up:
http://hl7.org/fhir/R4/procedure.html#bnr
Procedures are actions that are intended to result in a physical or mental change to or for the subject (e.g. surgery, physiotherapy, training, counseling)
Bob Dolin (Jan 03 2022 at 20:59):
Yes, agreed the definition of procedure could stand a bit of revision, because on the same page, we see that 'Examples include ... diagnostic procedures'
Kevin Power (Jan 03 2022 at 21:01):
Saw that, but then I also saw this:
The Procedure record is only necessary when there is a need to capture information about the physical intervention that was performed to capture the diagnostic information (e.g. anesthetic, incision, scope size, etc.)
Bret H (Jan 03 2022 at 21:01):
(was just going to paste "Note that many diagnostic processes are procedures that generate Observations and DiagnosticReports. In many cases, such an observation does not require an explicit representation of the procedure used to create the observation, but where there are details of interest about how the diagnostic procedure was performed, the Procedure resource is used to describe the activity.") Also, on the practical side, Bob had information about EPIC implementation of Procedure that made me think there won't be a conflict with hospital based processes....O and O is our goto.
Kevin Power (Jan 03 2022 at 21:01):
Either way, I am out of my depth and look forward to someone smarter than I to decide :smile:
Bret H (Jan 03 2022 at 21:03):
A big kicker is the event instance meaning for Procedure over Plan definition (which is defined to be definitional).
Bret H (Jan 03 2022 at 21:07):
if we shouldn't/can't create a new Resource then Procedure is a reasonable package. Importantly, it would be a profile off of Procedure - which gives a little distinction. The sad fact is that even if we created a new resource, it would look nearly identical to Procedure. We've determined to shy away from creating new Resources in the past.
Kevin Power (Jan 03 2022 at 21:08):
Doesn't look like Procedure to me if we are declaring we won't use 1/2 of the attributes :smile:
Bret H (Jan 03 2022 at 21:09):
Yeah, more like a constrained profile. you're right.
Bret H (Jan 03 2022 at 21:10):
take a look here sometime: http://build.fhir.org/workflow.html#event
Bret H (Jan 03 2022 at 21:11):
http://build.fhir.org/event.html
Kevin Power (Jan 03 2022 at 21:13):
Still feels like we said "this is an event, so which resource is the closest to what we want that follows that pattern" and not "yes, a profile of Procedure really fits our use case"
Kevin Power (Jan 03 2022 at 21:17):
But perhaps I am wrong - and fine by me if I am, great to have an answer for this and always good to learn something new :smile:
Bret H (Jan 04 2022 at 14:58):
A profile of Procedure as a process which is carried out fits our use case, without creation of a new resource.
The question went to O and O. They will provide a broader lab perspective which will be informative.
Jamie Jones (Mar 10 2022 at 16:22):
To update this audience, work has been done to draft a logical model of a new resource based on ImagingStudy. I've set up a working doc to help gather thoughts on how we may answer the questions in the "new resource proposal" form here: https://docs.google.com/document/d/1GCxY1mMpolqcyoNuCuUmFq0ribr3WSm5LiQ5TY_bGd8/edit# We welcome any and all comments!
Bob Dolin (Mar 10 2022 at 16:49):
Hi @Jamie Jones , we can potentially draw some content (e.g. use cases) from the earlier draft proposal https://docs.google.com/document/d/1N9JsMW3OXVO1VVipIYs-Cz6J0zpWEx51yvigMGlINOo/edit
Patrick Werner (Mar 15 2022 at 14:06):
Hi all,
to me it is still unclear why this has to be a new resource or has the working title "new resource proposal".
I think the created information model can be mapped to (profiled) existing FHIR resources.
A new resource also won't solve issues in R4.
Jamie Jones (Mar 15 2022 at 14:11):
My hope is that us filling out the new resource proposal document will help clear this up. Name should be GenomicStudy (mirroring ImagingStudy).
A lot of the first fields map easily but as you get further into the analysis/assay fields it diverges quite a bit. Example instances are sorely needed!
Bret H (Mar 21 2022 at 12:36):
@Jamie Jones there are use cases mentioned in the original draft proposal.
Bob Dolin said:
Hi Jamie Jones , we can potentially draw some content (e.g. use cases) from the earlier draft proposal https://docs.google.com/document/d/1N9JsMW3OXVO1VVipIYs-Cz6J0zpWEx51yvigMGlINOo/edit
Bret H (Mar 21 2022 at 12:37):
@Jamie Jones Who are you looking to that will provide example instances?
Jamie Jones (Mar 21 2022 at 12:39):
@Bret H I am happy to do a mapping exercise if we can get some GenomicStudy examples in another format
Aly Khalifa (Mar 22 2022 at 16:28):
@Kevin Power @Jamie Jones - regarding the GenomicStudy and the documentation/notes needed to push it forward, what are they? I'm trying to list them here up to my understanding and please add or correct me if I'm wrong.
- IG page content: Scope and usage, boundaries and relationships, and implementation notes
- Examples
- Use cases
Something else?
Kevin Power (Mar 22 2022 at 17:00):
Yes! to all the above :smile: (Well, not 'IG' page content, as a new resource will go directly into the main FHIR release, not our IG)
To see what all is expected, I would review the kinds of documentation and examples for FHIR resources today, such as:
http://build.fhir.org/imagingstudy.html (since we are basing off this)
http://build.fhir.org/procedure.html
http://build.fhir.org/observation.html
http://build.fhir.org/diagnosticreport.html
If you review those, I think you will get a sense of the sorts of documentation required.
Kevin Power (Mar 22 2022 at 17:07):
And feel free to start drafting that documentation via a Google Doc or whatever format you want. Someone else can convert it to the appropriate format to get it into the FHIR Core Build.
Kevin Power (Mar 22 2022 at 17:09):
The actual content is always the hardest part.
Jamie Jones (Mar 22 2022 at 19:11):
@Aly Khalifa I agree that thinking about the type of narrative on other resources is a great place to start, but would recommend you also focus on the specific questions we need to answer pulled in this doc: https://docs.google.com/document/d/1GCxY1mMpolqcyoNuCuUmFq0ribr3WSm5LiQ5TY_bGd8/edit . We have to convince FMG of those answers to begin the process of getting it into R5.
Kevin Power (Mar 22 2022 at 19:14):
@Jamie Jones - We did hear from Lloyd today that we are actually free to add the resource to the build anytime, but that FMG then reserves the right to kick it out (before R5 is published). So, we still do need that 'new resources' document filled out, but we are also free to add the resource to the build anytime someone is willing to do it :smile:
Kevin Power (Mar 22 2022 at 19:17):
But you are still right, we need to develop answers to the questions in that Google Doc as well. But perhaps right now, we can get the skeleton of the resource into the build?
Aly Khalifa (Mar 28 2022 at 12:41):
@Kevin Power - I started a document to include information relevant to GenomicStudy such as scope and usage.
It is available on https://github.com/khalifa-aly/HL7-CG-Models/blob/main/input/pagecontent/GenomicStudy.md
We may keep adding to it. I'll try to add this content to StructureDefinition-GenomicStudy.html later.
Kevin Power (Mar 28 2022 at 13:23):
Thanks @Aly Khalifa I will try to review today, but of course welcome others to review as well.
I will say do not worry about adding it to any HTML pages. When we add this to the FHIR Core, this content will be split out into a different file anyway. So feel free to focus on the content of this for now, and someone can work with it as it is being added.
Last updated: Apr 12 2022 at 19:14 UTC
