Stream: media issue
Topic: My Mash up of Option 2 for Discussion
Eric Haas (Oct 06 2017 at 23:21):
Observation
- remove valueAttachment ( so no longer two nearly identical ways to represent media based Observations)
- add element that references Media (so related Media Observations can be reference directly from Observation)
Media = Attachment + clinical context
- Keep as only way to represent media based Observations
- Move
height,width,framesanddurationfrom Media to the Attachment datatype ( they should be with the thing they describe) - Change basedOn to ref(Procedure)
Attachment = actual media content
- I think is likely no change in most (if not all) places Attachment is used instead of Reference(Media) -e.g. Patient Photo doesn't need to be more than an Attachment datatype
- Move
height,width,framesanddurationfrom Media to the Attachment datatype (they should be with the thing they describe)
As an aside in response to a suggestion to add Reference(Binary) I think the relationship between Binary and Attachment and Media and DocRef needs to be clarified see GF#14008
Eric Haas (Oct 06 2017 at 23:26):
Oh and under the Media heading
- Clarify the boundaries between Media and DocumentReference
Lloyd McKenzie (Oct 07 2017 at 05:03):
I don't understand moving height/width/frame/etc. to Attachment. Attachment is used for lots of things that aren't images. I don't see the point of moving them from Media if Media has 1..1 Attachment.
Eric Haas (Oct 09 2017 at 19:02):
I don't understand moving height/width/frame/etc. to Attachment. Attachment is used for lots of things that aren't images. I don't see the point of moving them from Media if Media has 1..1 Attachment.
There are two things that I find incongruous in Media adn one is those 4 attributes that are direct attributes of the content itself and not the context.
To me it is like putting Units of Measure in Observation instead of in the Quantity data type. They are direct attributes of the thing Itself if you don't need them you don't use them. I would be interested in how its implement in other domains.
John Moehrke (Oct 10 2017 at 13:35):
Eric, if Media is eliminated... do you then see the same problem with DocumentReference? I ask because I don't see any distinction between Media and DocumentReference. This is why I brought DocumentReference into the discussion...
Eric Haas (Oct 10 2017 at 15:11):
When I compare them I see a major difference in the context DR contains a lot more metadata about the the document ("document centered context") and media is more about the clinical context. I don't see Document reference as being a substitute for Observation like Media could be.
John Moehrke (Oct 10 2017 at 15:14):
Okay, then I agree. If we didn't have Media, then DocumentReference would then be more clear. (I might still recommend a rename of DocumentReference to be more clear that it is not limited to "CDA Documents". Which is a current interpretation that has never intended to be true).
Elliot Silver (Oct 10 2017 at 17:34):
@Lloyd McKenzie , if we define Media as being used for "clinical multimedia," and not for general purpose images/video/audio, then most other images/video/audio will be represented with an attachment. For these uses, width/height/duration/frames are useful. Therefore, these attributes should be added to Attachment.
I recognize that there are other attributes that might be useful for other content types, and that there are slippery slope and 80% issues to be discussed, but I think idea is valid.
Lloyd McKenzie (Oct 10 2017 at 18:55):
We would end up with page-count, MS-Word version and a whole lot of other stuff. If we were going to have them on attachment, I'd definitely lean toward extensions.
Eric Haas (Oct 10 2017 at 22:02):
size covers page count and ms_word version sounds like as extension to me :-)
John Moehrke (Oct 11 2017 at 13:37):
actually MS-WORD version is already a mime-type...
Eric Haas (Oct 23 2017 at 07:16):
Here is a visual of what I am proposing using clinFHIR excellent tools - This address the Media issues as well as GF#10118 (related types) and GF#14034, GF#13347 simply and elegantly IMHO but constitutes a couple of major breaking changes to Observation.
Eric Haas (Oct 23 2017 at 07:17):
(deleted)
Eric Haas (Oct 23 2017 at 07:17):
Screen-Shot-2017-10-23-at-12.10.35-AM.png
Eric Haas (Oct 23 2017 at 07:18):
(deleted)
Eric Haas (Oct 23 2017 at 07:19):
Screen-Shot-2017-10-23-at-12.10.05-AM.png
Eric Haas (Oct 23 2017 at 07:21):
can check it out here: http://clinfhir.com/logicalModeller.html#tti59
John Moehrke (Oct 23 2017 at 15:09):
so it doesn't have Attachment datatype within Observation? Seems this would work, with a recomendation that when this is needed a contained Media can be used... right?
Eric Haas (Oct 23 2017 at 17:07):
yes removed value-Attachment. The recommendation is to use Media directly instead of Observation.
Not shown is the second part of my proposal: slim down Media moving the image details to Attachment Datatype
And the third part of my proposal is let all the WGs decide whether to use Attachment vs Media vs DR for the other resources that use Attachment.
Last updated: Apr 12 2022 at 19:14 UTC
