Stream: ihe
Topic: IHE Workflow conformance
Jose Costa Teixeira (May 16 2018 at 09:34):
Hi. Looking at using planDefinition as a placeholder as a workflow profile (in the IHE sense).
Jose Costa Teixeira (May 16 2018 at 09:36):
pinging @Bryn Rhodes ,
for planDefinition to support our needs, we need at least the following:
Jose Costa Teixeira (May 16 2018 at 09:39):
1. associate a conformance resourceto a given step (e.g. in step 1, the resource must support structureDefinition1)
2. support "machine" actors besides human actors.
...?
Jose Costa Teixeira (May 16 2018 at 09:40):
2 would be an easy change to participant.type
Jose Costa Teixeira (May 16 2018 at 09:40):
1 i have no idea.
Jose Costa Teixeira (May 16 2018 at 09:41):
If we solve those, I think we could start doing some work and then figure out the rest.
Bryn Rhodes (May 16 2018 at 10:03):
@Jose Costa Teixeira , I suggest we start with an extension that would allow us to set the profile on PlanDefinition.action.definition for 1. For 2, I'm actually kind of surprised we don't have the ability to specify that a participant is a device, so I think we'd be happy to expand that value set. If you agree with both of these, please submit trackers.
Jose Costa Teixeira (May 16 2018 at 10:05):
aah i missed PlanDefinition.action.definition. Sounds good.
Jose Costa Teixeira (May 16 2018 at 10:16):
#17200 and #17201 submitted
Jose Costa Teixeira (Apr 20 2020 at 19:09):
Getting back to this,
Jose Costa Teixeira (Apr 20 2020 at 19:11):
discussion on how to express "workflow conformance" in the sense of an IHE transaction - like: "actor 1 sends this resource with this profile, actor 2 responds with this profile"
Jose Costa Teixeira (Apr 20 2020 at 19:13):
We have discussed so far that PlanDefinition could support that.
I started putting something together, taking a rather simple set of activities (in this case, IHE MMA transaction for getting the medication administration worklist for a nurse), and materializing it here (with some open questions)
https://confluence.hl7.org/display/FHIRI/Workflow+Conformance+example
Jose Costa Teixeira (Apr 20 2020 at 19:14):
I'm reopening this so tat we are all on the same page so that we work efficiently when implementing.
Jose Costa Teixeira (Apr 20 2020 at 19:15):
Initially I hesitated with using PlanDefinition for that, but I've been reassured that this would be a good option. Perhaps others want to chime in.
Jose Costa Teixeira (Apr 20 2020 at 19:16):
@Bryn Rhodes @Keith Boone @John Moehrke @Elliot Silver , others
Keith Boone (Apr 21 2020 at 08:18):
In order to model sequencing, I found it necessary to have a way to recurse w/ actions. Actions beget other actions, which eventually complete and return a result to the caller... some are synchronous (modeled recursively), others asynchronous. PlanDefinition doesn't really have the capacity too support that in a way that was easy to process via a transform. The information could be captured, and the recursive structure regenerated, but the complexity of representation just didn't match the way people think. Accurate representation is good, but making it easy for people to express things the way THEY think rather than the way the computer thinks is really important. We switch context a lot when we describe what happens. The XYZ sends message 1 to ABC, which then (context switched to ABC) sends message 2 to DEF. When it (context switched to DEF) returns (back to ABC), then a result is passed back (popping back to XYZ) to XYZ.
Gotta get that sense in the resource, and in a way that is natural for the user.
John Moehrke (Apr 21 2020 at 12:32):
cucumber -- modern language that aligns well with established layout of IHE Volume 2 (trigger event, message encoding, expected action ) 1..*
John Moehrke (Apr 21 2020 at 12:33):
where a trigger can be the completion of previous, completion of async X event, failure-modes, timing, user input, etc...
Jose Costa Teixeira (Apr 21 2020 at 13:11):
is cucumber an artefact?
Jose Costa Teixeira (Apr 21 2020 at 13:11):
or can it be transformed in one? (or embedded in one)?
Jose Costa Teixeira (Apr 21 2020 at 13:11):
(with an id, status, owner - to make it manageable)
John Moehrke (Apr 21 2020 at 13:12):
There are two IHE profiles that are experimenting with using Cucumber in an appendix to explain the expectations for testability.
But, my point is that Volume 2 is already laid out in this pattern. So it is not invention
Jose Costa Teixeira (Apr 21 2020 at 13:14):
what is the artefact that we should use then?
Bryn Rhodes (Apr 22 2020 at 18:18):
@Keith Boone , I'd be interested to see examples of workflows that aren't well representable, would be good feedback to the design of the resource. PlanDefinition does have nested actions, so at first glance I'm not seeing where the gaps are.
Keith Boone (Apr 22 2020 at 19:11):
Whoops, I didn't look far enough down the page... Which might be another challenge. PlanDefinition seems like it might be trying to meet the needs of many masters. I do have to spend some time looking at the details of it in more depth, rather than 30 second reviews of the content.
Last updated: Apr 12 2022 at 19:14 UTC
