The FHIR extension fields (defined in the common data model or Alvearie FHIR IG) have changed since the code was written
mcbride-p opened this issue · 6 comments
The back-end will need to change to use the new fields and so will the UI (and will have to coordinate the changes to not break demos, etc)
From Eric on the 1/10/2021 (copying Lisa Wellman, Kathryn Whalley, Nick Lawrence....)
Nick is working on updating the health-patterns copy of the code and making a number of changes, including these
Kathryn/Lisa have made these changes in the CDP code and in the testcases, etc.
Does anyone have a list of before/after names or list of changes in general?
If not, it may be good to look at a couple of output examples before & after. Our intern had some examples here:
https://ibm.ent.box.com/folder/143277372649?s=vmy5m4yv995z6iw8k7lsynjsfhec0rrd
and things like Immunization and Condition and DiagnosticReport would have the old values that she used (and got from the CDP team originally)
CDP will have updated these to the new names so at worst a comparison between old and new could show you the meaningful changes.
Any other ideas ?
Lisa's response on the 1/10/2021 ....
It was not just URL renames, but changes in the nesting.
This is the commit for our work if that helps https://github.ibm.com/WH-Provider-Analytics/whpa-cdp-lib-fhir-enrichment/commit/559824551a8075fab41657a64c4392319cc4a0c6#diff-7ea13304258a6c5559af432b6b0d5d66
Nick's response on the 2/10/2021 ....
For the changes I am making, I think we will want an extensive review at some point. A big part of that is the common data model, and the second big part of that is the changes to the code to work with QuickUMLS and be open source.
I pulled a few weeks ago from CDP and have been using that as a reference. I noticed a couple of days ago that there had been changes made since I pulled. I haven't worried about it much because I have my head down getting the existing code updated for QuickUMLS. Any new structural stuff should be easier to address once things are working at the commit level I started with.
The code changes are primarily focused on the need to share code between NLP engines, but there are significant flow changes to accomplish that.
There are also changes for source & quality being made at the same time.
- Apache License copyrights
- PEP8 compliance (black)
- Correct docstrings and add doctests where appropriate
- Use of type hits (PEP484)
- correction of pylint, flake8, mypy reported errors and warnings
- method design and data type enhancements
Given that I'm only supposed to be part time on this project, I think I'm about a week or two from having it ready for review. (The unit tests I got from CDP are working, but I still have yet to make the equivalent changes for QuickUMLS, and plug the whole thing into the NLP service. Ideally we'd write some QuickUmls unit tests as well).
Once it's ready we'll probably want to have a couple of high level reviews. We need to make sure the data structure of the FHIR resources is correct (unit tests help greatly with that!). And we also need to understand the changes better before we can talk about how to better collaborate in the future.
And then from Erik on the 4/10/2021 ...
Paul - given the timing that Nick outlines below, it may not be worthwhile for you to look into this further during this sprint. We will talk to see if there is anything we can do to accelerate getting the updates but I wouldn't want to have the UI complete but the back-end not able to create the FHIR structures that you need.
Putting this on hold at least for current sprint 29.