Stream: ihe
Topic: DocumentManifest v Provenance
René Spronk (Jul 01 2019 at 09:46):
Why is there a separate DocumentManifest when the use case in XDS for SubmissionSets is effectively reduced to that of a Provenance resource ? Couldn't we just use Provenance instead?
John Moehrke (Jul 01 2019 at 14:55):
Interesting point.. Might be very useful to look into this. DocumentManifest is not a well loved resource...
John Moehrke (Jul 02 2019 at 15:42):
I looked at this a bit. Although high-level functionally in an XDS environment a SubmissionSet is pretty close to high-level functionally a Provenance; there are some elements that are less convenient in a Provenance resource, such as the sender (author of the SubmissionSet, is not necessarily the author of the DocumentReference or documents), recipient, and reason for the package. Most of the data in a SubmissionSet can be placed into a Provenance, but much of it must be encoded two layers deep in Provenance.agent elements. This is possible, but makes for Query difficulty. Most dissonant is that a SubmissionSet has a lifecycle itself, where there is no lifecycle of a Provenance (i.e. there is no Provenance.status, and should NOT be). --> I don't think Provenance is a good fit for SubmissionSet.
René Spronk (Jul 03 2019 at 06:49):
Ah, but shouldn't we be looking at how SubmissionSets are actually used in XDS projects, instead of just looking at the XDS spec itself? Especially when it comes to SubmissionSets (and Folders) there is a large discrepancy between the theory (from 2005) and the practical use (today) - we do have to cover the 80%, the rest can be an extension.
Who would ever query for a submission set? IMHO only some old radiology systems that don's support ReferenceId list, on the assumption that the images and the report were submitted in one single submission set.
The life cycle of a submission set: probably the most problematic issue when one compares a submission set to a provenance on a one to one basis. But the use case (state changes on a submission set) is exotic, each state change could perhaps be modeled as a separate Provenance. I agree that a provenance itself can't have state changes.
John Moehrke (Jul 03 2019 at 11:49):
I think I would rather see it become a flavor of List than a flavor of Provenance.
René Spronk (Jul 03 2019 at 14:44):
Given that grouping mechanism is the only use case I know of (for SubmissionSet), seems worth exploring.
Last updated: Apr 12 2022 at 19:14 UTC
