Stu2 ValueSet updates

[Dtphelan1]

Summary

Updates ValueSets(VS) to use STU2 equivalents. This PR also adds a markdown file and node-script, explaining and enabling the VS expansion pipeline for ease of future use. Since this expansion is scripted, a number of VS's beyond what we use today were generated; all VS based in the elixhauser IG were removed and all VS stemming from mCODE were left in, even if unused.

Below, I highlight VS's we used , and what changes have been made.

• ValueSet-mcode-cancer-related-radiation-procedure-vs.json
  • Valueset was removed and replaced with several other radiotherapy valuesets
• ValueSet-mcode-cancer-related-surgical-procedure-vs.json
  • Codes were removed
• ValueSet-mcode-cancer-staging-system-vs.json
  • Codes were added
• ValueSet-mcode-primary-or-uncertain-behavior-cancer-disorder-vs.json
  • Codes were updated
  • Codes were added
• ValueSet-mcode-secondary-cancer-disorder-vs.json
  • Codes were removed
• ValueSet-mcode-tumor-marker-test-vs.json
  • Codes were removed
  • Codes were added

Of note: Several expansions failed, and instead of a VS with an expansion field we have the original compose field, intensionally defining the VS. Importantly, this happens to our cancer-related-surgical-procedure VS to the point of unusability. We need to decide if we're okay with a janky Surgial-procedure extractor for now, or if we want to retain the old, more-usable VS to maintain Extractor functionality; while we discuss I've left the test broken.

Below is a full list of VS that fail expansion, with bolding applied to those VS we've had prior to this PR:

• ValueSet-mcode-primary-or-uncertain-behavior-cancer-disorder-vs.json
• ValueSet-mcode-radiotherapy-modality-vs.json
• ValueSet-mcode-radiotherapy-technique-vs.json
• ValueSet-mcode-radiotherapy-treatment-location-qualifier-vs.json
• ValueSet-mcode-radiotherapy-treatment-location-vs.json
• ValueSet-mcode-secondary-cancer-disorder-vs.json
• ValueSet-mcode-tumor-marker-test-vs.json

New behavior

No new code behavior in the functional MEF; new behavior is restricted only to what codes are semantically tied to certain mCODE concepts. There is a new script for auto-expanding future VS – vs-expansion-script.js.

Code changes

New vs-expansion-script should be reviewed.

Testing guidance

• Ensure old tests still pass.
• Discuss and resolve open questions re: un-expandable VS.

[Dtphelan1]

The recent torrent of commits is the result of Julia, Julian, and me collaborating in the WS. With her careful feedback, I've successfully:

1. used the correct version of the spec;
2. streamlined the automated-expansion process;
3. manually expanded several VS, including ;
   • secondary-cancer-disorder;
   • primary-or-uncertain-disorder;
   • surgical-procedure.
4. updated tests to resolve failures.

We may still want to take a more sensible approach to 4, so I'm open to feedback.

[jafeltra]

I had a few comments inline, but I had one other question in general:

• I think the Tumor Marker Test VS and Surgical Procedure VS were both ones we manually expanded, correct? Do you think there's a need to include the unexpanded filters in our ValueSets? I haven't actually looked into if there's a valid way to include both the expanded codes and additional filters in the FHIR resource, but I can if that's helpful in answering this question.

[jafeltra]

• I think the Tumor Marker Test VS and Surgical Procedure VS were both ones we manually expanded, correct? Do you think there's a need to include the unexpanded filters in our ValueSets? I haven't actually looked into if there's a valid way to include both the expanded codes and additional filters in the FHIR resource, but I can if that's helpful in answering this question.

I talked to Dylan about this, and he reminded me that our logic for checking codes in value sets only looks in the expansion or the compose so there's nothing we should update in our ValueSets to include both, since both won't be checked.

[jafeltra]

One last thought after your most recent changes: should we remove all the radiation procedure related value sets since we removed the extractor entirely?

[jafeltra]

Looks good to me!