Stream: FHIRcast
Topic: new-event-template create and update
Bill Wallace (Sep 12 2019 at 17:57):
It would be useful to have events for creation of a resources such as DiagnosticReport. Something like:
diagnosticreport-create
to indicate that the given component is starting to create such an object
Then,
diagnosticreport-update
to indicate that the given changes apply to the resource. The way this would be used is to include the sections that are being modified, although if some part of hte text is changed, it may require include the entire body of the report changes. That suggests only occasionally updating the report body
Similarly
imagingstudy-update
could be used to send markup changes that are in-progress, by including (private for now) elements defining the markup. That would only need to be sent on markup which is changing for the images it applies to.
George Kustas (Sep 12 2019 at 19:42):
@Bill Wallace , it is very coincidental that you posted this today since I just had a similar conversation with another vendor. We were discussing the FHIRCast integration of a reporting system and PACS system, and how the two systems frequently work on a "case" involving multiple images. It is called "accession linking" or "combined results" (in HL7 terms). The PACS usually controls the work list in this environment, and decides which imaging studies to include in the case (diagnostic report). With existing integrations, the PACS would tell us (the reporting system) to create a new report, and use accession numbers [..., ..., ...]. In this environment, we're thinking it might be appropriate to use a DiagnosticReport-Open event, which contains one or more ImagingStudy resources in its payload. By the way, the PACS may want to include old, prior studies for comparison. We are implementing a prototype that will allow for such a thing by using the "key" tag and specifying a value of "PriorStudy" as opposed to "Study". Of course DiagnosticReport-Update would naturally follow allowing similar changes to accessions, addition of other things that you mention (like report contents), and possibly other pieces of structured data that could be contained in an Observation resource (like findings, measurements, etc..
).
Would love to hear comments or opinions from anyone on this subject. Maybe STU2 @Isaac Vetter ?
Bill Wallace (Sep 13 2019 at 17:09):
Bill Wallace , it is very coincidental that you posted this today since I just had a similar conversation with another vendor. We were discussing the FHIRCast integration of a reporting system and PACS system, and how the two systems frequently work on a "case" involving multiple images. It is called "accession linking" or "combined results" (in HL7 terms). The PACS usually controls the work list in this environment, and decides which imaging studies to include in the case (diagnostic report). With existing integrations, the PACS would tell us (the reporting system) to create a new report, and use accession numbers [..., ..., ...]. In this environment, we're thinking it might be appropriate to use a DiagnosticReport-Open event, which contains one or more ImagingStudy resources in its payload. By the way, the PACS may want to include old, prior studies for comparison. We are implementing a prototype that will allow for such a thing by using the "key" tag and specifying a value of "PriorStudy" as opposed to "Study". Of course DiagnosticReport-Update would naturally follow allowing similar changes to accessions, addition of other things that you mention (like report contents), and possibly other pieces of structured data that could be contained in an Observation resource (like findings, measurements, etc..
).
Would love to hear comments or opinions from anyone on this subject. Maybe STU2 Isaac Vetter ?
My feeling is that the open event should be for starting to view an existing object, whereas create and update are used for initial setup and subsequent modifications. It partly depends on how much one wants to overload the open operation. I guess I'm worried about systems that try to match open/close counts exactly.
George Kustas (Sep 15 2019 at 19:33):
@Bill Wallace I posted a couple of rough proposals on the fhircast-docs page: https://github.com/HL7/fhircast-docs/issues/289 and https://github.com/HL7/fhircast-docs/issues/290. I participated remotely in the connectathon this weekend and discussed these briefly. It will be great to have another vendor developer like you involved.
Last updated: Apr 12 2022 at 19:14 UTC
