Skip to end of metadata
Go to start of metadata

Date: Monday, September 16, 2019

Quarter: Q3

Live doc for notes:

FHIR Connectathon review

Goal: QA+Examples (+other remaining gForge trackers)


Minimal examples can validate but not be a meaningful representation of a use case.

Need to finish collating meaningful examples and queries.

IG Validation

TBD Code reconciliation

Some concepts may still overlap/align with available codes.

Some are currently being ironed out with LOINC.

Terminology works in progress

  1. GL-string support for HAPI
  2. Sequence Ontology for ontoserver / CSIRO a new/upcoming open source OWL - FHIR code system project
  3. - we currently require “HGNC:[id]” for gene ids, format is “[id]” for gene families. Should be able to represent gene families as a concept, preferably at the report level.
    1. Aim to create a separate code system for gene families

Next steps:

File trackers for missing examples and queries SPECIFICALLY with meaningful titles and descriptions (to enter before we leave Atlanta)

Future topic (sometime this WGM)

Implementation concerns in mapping between models potentially starting to adopt DTSU2 (epic)

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."


FHIR Connectathon review

Prep for joint with FHIR-I

Discussion items


10minFHIR Connectathon review

1 hourPrep for meeting with  FHIR-I

Action items

Quarter: Q4

Live doc for notes:

Questions for FHIR-I

New questions:

  1. Codesystem creation / namespaces hgnc gene families

Problem: HGNC has overlapping number ranges for genes and gene families (genes use the prefix HGNC:,

In terms of gene IDs, not approved symbols or names!/group/588


Just “588” is what is found in the wild but incorrect, i.e., implementers must add the “HGNC:” prefix to find the right concept. However, anyone attempting to retrieve gene families will then be routed to an incorrect gene.

families don’t have a prefix. As prefixes could be forgotten there is a danger to confuse genes and families)

HGNC doesn’t think that is a problem and aren’t into creating separate FHIR CodeSystems, or any FHIR CS for the both concepts.

Are we allowed to create CodeSystems in foreign NamingSystems?


Can we define CodeSystems in their namespace after attempting to collaborate?

A: we’ve never had a complaint about it yet!

Potential proposal:

Create separate codesystem for gene families and gene ids inside their namespace

Deprecate? The old codesystem

Get in touch with HGNC, if they aren’t willing to collaborate we will do it for them

  1. How to define/measure conformity to our IG
    1. Advertising:
      1. CapabilityStatement can “instantiate” other capabilityStatements
      2. Can also point to an ImplementationGuide
    2. Proving support
      1. Ability to execute (Aegis/crucible)
      2. Most powerful would be a hosted service for this, but difficult
      1. TestScripts?
      2. Build a small reference client and perform round-tripping with 
    1. Packages inside capabilitystatements
    1. You can define them per profile, and for each element, in prose
    1. Advertising vs conforming?
    2. Are all profiles necessary?
  2. Artifacts currently in R4
    1. Can ask to annotate remaining pages with disclaimers (some have)
    2. Can remove some artifacts in current build (so will be taken out for R5)
    1. Grahame putting together a pile of items for technical corrections after the WGM
  3. Tooling errors in the build process
    1. Continue doing that
    1. How best to follow-up (current methods: poking grahame and filing trackers)?
  4. Trifolia timeline
    1. “A way’s out, if it comes”
    2. IG templating: mapping from current method
  5. R5 timeline
    1. Draft ballot aimed at May (delayed from February)
    2. Core clinical resources should be normative by R5
  6. Knowledge resources
    1. CDS workgroup stuff
      1. Plan definition / library/ clinical quality
    1. In general, how are other groups progressing?
    2. Definitional resources?
    3. Compare to our Implications
  7. Open floor
    1. IGs are hard to read, conformance resources aren’t being utilized
      1. Establish authoring patterns, focus on rendering and breaking validation down from On/Off
    1. What do you consider the biggest gap in FHIR infrastructure?
    2. What molecular labs are currently using FHIR?

Date: Tuesday, September 17, 2019

Live doc for notes:

Quarter: Q1

presiding co-chair - Bob Milius
scribe - Bob Freimuth

  • Review Agenda
  • Introduction of attendees
    • Nearly full room (18) - may want to consider a larger room next time
    • Review Agenda for WGM
    • Review of CG for newcomers
      • HL7_CG_intro_Sep2019.pptx
      • Add CCDA family history to the list of products (per Keith Boone)
      • Brief review of Genomics Reporting IG
      • Last balloted version, Jan 2019
      • Emphasis on consistent, computable implementations; we must all implement the same way or interoperability will suffer
      • Profiles derived from other Profiles (slide 24)
        • This slide illustrates some of the important relationships between core observations within the genomics reporting IG
        • Note to co-chairs: increase size of images; many of the examples on slides 26+ are too small for people to read from the back of the room
      • Slide 33: example of adding a coding that might be used by the lab for reporting, but which is not required by the IG spec
      • Slide 34: example of post-coordination using components
  • Discussion
    • Are we interacting with other groups, such as ISO?
      • Yes, we have regular “external efforts” updates on Tuesday calls, we rely on members to provide touch points to other groups
    • Keith: We should be working with the HL7 terminology authority (HTA) on issues relating to code systems; they can help ensure consistent use of vocabs across HL7 products
      • Relationship with vocab WG?
      • HTA is accountable to HL7 board and is responsible for coordinating with external groups that produce vocabs; establish agreement tor reuse content, IP/licensing, etc
      • HTA will coordinate business relationships and governance issues related to the use of vocabs; FHIR-I and Vocab WG will handle the tech implementation details
    • How do we find the latest version of the IG?
      • See also “version history” in the red banner at the footer of any IG page
    • Will we need to update the v2 LRI because of what we’re doing in the IG?
      • FHIR supports some things that cannot be easily done in v2; we should review the LRI in detail the next time it comes up for ballot and determine if we should continue to support it or not
      • Registry of all HL7 IGs:
      • How aligned is LRI with current IG?
        • Lloyd’s first attempt at the FHIR IG was based on the LRI, many of the core concepts should be retained in it but the mapping should be reviewed
      • Should we be participating in the “v2 to FHIR” effort?
        • Keith is participating in this, they have been doing some work with OBX, but they are not going into the level of detail needed to support the data in our reports
        • Keith: v2 to FHIR is building mapping tables (spreadsheets) that show how fields in OBX map to attributes in FHIR resources
        • Currently focusing on readability to facilitate implementations
      • Are concept maps unidirectional?
        • Keith: if maps are done correctly they can be bidirectional; must be very careful or might have information loss, which makes the mapping unidirectional
      • Use of coding systems appear to be aligned well between v2 and FHIR (in genomics)
    • Code system status (Patrick)
      • Current FHIR validator has a syntax issue, thinks URIs with/without trailing slashes are different, causes some links to code systems we use are invalid
      • SequenceOntology expansion
        • We are using concepts under SO sequence_alteration as a replacement for LOINC DNA change type
        • We need a way to expand a value set for SO into a FHIR code system resource
        • This is a limitation in current tooling, no automated way to do it (yet)
        • This is necessary to support value set expansions (concrete lists of concepts rather than simply a reference to a single parent concept)
        • Some proprietary tooling exists to convert an OWL ontology into a FHIR code system resource
        • Shrimp: (concept browser for FHIR code systems)
        • Patrick will look into a similar, open source toolset to expanding value sets automatically
      • HGNC
        • HGNC supports two different code systems but we currently have only a single URI, trying to handle it with namespaces and identifier syntax rather than just registering another URI
        • Suggestion to take 2 approaches in parallel: do a workaround now to address the patient safety issue, but also work with HGNC to encourage them to improve their approach to code systems and URIs (and identifier syntax)
        • Must work with HGNC to ensure they support our suggested namespaces and are willing to commit to their stability over time

Quarter: Q2

presiding co-chair - Patrick Werner
scribe - James Jones

  • Intro to GA4GH (Bob F)

  • Introduction to the phenopackets project. (Alejandro Metke, Bob F)

    • Slides:

    • Phenopackets

    • Structured, computable description of phenotype data

    • What was observed/not observed

    • Supervised mapping from HPO (human phenotype ontology)

    • phenotypes

    • Diseases are separate from phenotypes in the spec, as a disease here represents a diagnosis (has onset)

    • Pedigree (based on PED format)

    • Variant (can use HGVS, VCF, SPDI, ISCN notation) (will implement GA4GH VRS)

      • Nephi: why do phenopackets send genotype information?

      • Nephi: can they support provenance/generation of the phenotype?

    • Phenopackets on FHIR

      • Close ties with Australian Genomics and Genomics England

    • Requires some extensions in current approach

    • Multiple resources could be targets for mapping concepts,

      • Determine resources, define profiles & extensions

      • Composition for the packet itself, others used as well

    • Where should it live? What HL7 work groups should own an IG in this space?

      • CG / BR&R / others?

      • Several attendees supported CG as the natural “home” for this work, with co-sponsorship by BR&R and O&O

    • How to take the next step?

      • ~5 week timeline

      • Consider a PSS for this in the near future

  • How to support Implementers? (Arthur)

    • What is the approach external groups can take when

      • Needing information on how to use our model?

        • Likely outside the official scope of HL7

    • Suggesting changes to the spec

      • Anything faster than the Ballot process?

      • Keith: (comment from Lloyd) IGs should address both coders and business decision makers as target audiences

      • Swapna: major clinical labs are still very far behind

      • Bob F: do their contracted software developers have seats at these tables?

      • Bret: is there incentive for them?

      • Bob M: should we develop an accelerator program?

Quarter: Q3

presiding co-chair - Patrick Werner
scribe - James Jones

  • Alignment with mCode
    • Presentation about mcode and their genomics profiles by May Terry: 
    • mcode Variant-Test requested and Variant tested
      • Request: Does the patient have variant XY
      • Answer: yes/no
    • Discussion about positive/negative test reports and “what was looked at” : 
    • Bob M: Need guidance how to associate region studied with an obs-variant
    • Marc: Ideally there is a IG core IG/ simple IG which can be used by derived profiles
    • Patrick: we could do derived profiles/IGs which are only covering smaller sub use-cases
    • James J: will provide capability statements for specific use cases in the next release
    • May: SpecimenType is an Extension on the DiagnosticReport to avoid to create a Specimen
    • Avoid 0..0 restrictions if possible, otherwise Resources containing more information than needed are getting rejectd
    • Avoid Extension if it can be done via components as components are part of the core spec and have to be supported (extensions can be ignored), also Extensions need custom search queries whereas components are supported by default 
    • Complex Extensions lead to an “all or nothing” problem, if the system doesn’t support the Extension all information will be ignored. If you have atomic Extensions you won’t lose everything but the unknown extensions.
    • next steps: Continue mCODE gap analysis on CG FHIR Modeling meetings on Mondays.  If possible, post artifacts to review or questions to address ahead of time to 
  • We need a “zombie” icon for FHIR (indicates “deprecated”)

Quarter: Q4

presiding co-chair - Bob Milius
scribe - James Jones

  • gForge issues
  • TMB proposal
  • Questions for OO?

Date: Wednesday, September 18, 2019

Live doc for notes:

Quarter: Q1

presiding co-chair - Bob Freimuth
scribe - Bob Milius

  • Information Model

Quarter: Q2

presiding co-chair - Bob Freimuth
scribe - ?

  • Working meeting for Modeling - Tentative

  • separately, joint meeting with OO (they are hosting)

Quarter: Q3

presiding co-chair - Bob Freimuth
scribe - Bob Milius

  • Sep WGM review

  • CG business

  • Mission, Charter, SWOT, DMP

  • February WGM planning (Australia)

Quarter: Q4

presiding co-chair - Patrick Werner
scribe - Bob Milius

  • FHIR & IG summary