Skip to end of metadata
Go to start of metadata

Date: 2/6/2020


Quarter: 1


Minutes Approved as Presented 


This is to approve minutes via general consent. "You have received the minutes. Are there any corrections to the minutes? (pause) Hearing none, if there are no objections, the minutes are approved as printed."

Goals

Set goals, objectives or some context for this meeting.

Discussion items

TimeItemWhoNotes
  • Joint with CQI, CIMI
  • Chairing: Bryn Rhodes.  Scribe: Ken Kawamoto
  • Quality Improvement Specifications Roadmap
Bryn

Reviewed CDS standards roadmap from Clinical Decision Support Standards, which was taken from the HL7 wiki.

Discussed migration of the CDS Knowledge Artifact Specification to the FHIR-based Clinical Reasoning specification.


  • CIMI-based logical model
  • FHIR profiles for CDS and CQI
Claude, others

Discussed need for logical model that can handle transforms to/from various versions of FHIR, FHIR profiles, and vendor implementations of these which can be different.

Claude presented information from the American Medical Informatics Association (AMIA) 2019 Annual Symposium in 11/2019 on how QUICK was developed as a CIMI-based logical model for CDS and eCQM, aligned very closely with FHIR.  Current state:

  • Technical validation phase (done)
  • QI Core profile generation pilot (in progress)
  • Model development (in progress)
  • Mappings to FHIR DSTU2, STU3, R4 (in progress)
  • QI Core profile generation for FHIR DSTU2, STU3, R4
  • Field testing

Floyd raised a few issues:

  • If QDM goes away and we don't have QUICK, what should eCQM authors write against?
  • Concern about QUICK – someone has to write the tooling that will do those mappings.  Those individuals/groups must be able to keep it sustained, with sustained resourcing/funding.
  • Backward compatibility – how to include items that are in newer versions of FHIR or FHIR profiles but would only be expressable as extensions in the source FHIR implementations.
  • Discussed that CMS may not be the right long-term funder; it may need to be a group like ONC.  Need to have a conversation with ONC and CMS, and a roadmap.
  • QDM not going to be supported with FHIR R5.  May be looking at QICore.

Bryn brought up FSH (FHIR Short Hand), a DSL for Structure Definition.  A language for expressing a model.  Could help with maintaining QI Core moving forward.

Isaac brought up normative FHIR resources as the way forward; perhaps this will not be an issue once we get there.

This is a generic problem across use cases beyond CDS and eCQM.

Discussed: where do we work on the complexity – at the level of the knowledge artifact, or at the level of the logical model?

Logical model(s) could be particularly useful when mapped to a FHIR Profile level (e.g., QI Core).

3 levels where this can be done (perhaps all 3 are needed):

  • Logical/conceptual model
  • Common functional libraries (e.g., to get effective time)
  • Core logic

Someone needs to do this hard work.

Feedback from vendor: we need to explain why this is needed, in a way that vendors and payors understand.


  • Personalized Decision Support via CDS Hooks (Adaptive Cards, presentation from Muhammad Afzal w/ South Korean UCLab Kyung Hee University/Sejong University)

Muhammad

See presentation –

Discussion:

  • Card UI is intended to be governed by the EHR client, not the CDS Hooks service.  So even though a SMART on FHIR launch allows UI theming information to be sent to the app provider on launch (e.g., low-light, large fonts, etc.), discussed that this is not something that we plan to provide in the CDS Hooks launch context.
  • Data such as mental health information, physical activity, diet pattern are already covered in scope via other FHIR modeling activities

  • QI-Core FHIR Trackers

Deferred due to time

Action items


  • White paper on business case for a logical model and transform tooling.  A part of transition to FHIR-based quality measurement.  There is a PSS to support this.