Minutes Data Working Group 28 Aug 2020

Agenda

• Review FHIR encoding of device observations
• Review FHIR tools
• Check in on MONAI development needs re: Data Working group
• Next steps

Minutes

• Discussion on FHIR Objects
  • Question: Can other mechanisms be used within FHIR to represent data? E.g., as a table structure?
    • Not really. If other mechanisms are used (for example, SampledData), the interoperability mechanism would be more difficult
  • The data in this chart is not representative of data typically clinically represented (e.g., state of the lights in the OR room)
  • Concern that the data structure is not "accessible" to the general researcher
  • Could YAML be used instead of FHIR? Engineering discussion suggested looking at YAML
    • Any YAML stucture would be "proprietary" in nature, which means that transformations would be required regardless to fit into the structure
    • There is a library for FHIR-YAML representation
• Discussion on FHIR tools
  • Focus should be on python libraries - HAPI and Dot Net libraries, while interesting, not as applicable to engineering work now
  • SMART on FHIR Python libraries looks like the right fit - but are they reusable outside of a client-server architecture?
    • Need right now is not on a "client-server relationship" of FHIR, but more of "memory representation) of FHIR
    • FHIR Parser might be useful for handling proprietary FHIR extensions
    • Brad to reach out to project and get more information / ask if interested in joining for a chat
• Engineering needs 0.3 for Data Working Group
  • Plan is to have a working prototype of standardization - e.g., a simple FHIR parser - in memory storage
  • Would a simple image classification representation be represented by FHIR that wraps DICOM pointers (ImagingStudy? NIFTI?)
    • What does the coding/ontology aspect look like? Could you use observation and have it relate to the imaging data?
  • IXI dataset looks like an interesting set to model in FHIR
    • e.g. could there be a way to convert the CSV files to FHIR?
• Action items
  • Brad to reach out to SMART on FHIR Python group for information / discussion around representation in FHIR
  • Engineering team can already start looking at the SMART on FHIR parser; Wenqi to report back on how that's going on next call
  • Study the IXI format and consider modeling for FHIR resources of what representation could look like