2017-06-15 CIMI Telecom Minutes

* 2017 Minutes: http://wiki.hl7.org/index.php?title=CIMI_Minutes 

• Screen Sharing & Telecom Information: IHTSDO has generously provided a GoToMeeting connection for CIMI use.
  • Please join the meeting from your computer, tablet or smartphone at https://snomed.zoom.us/my/snomedhl7
  • Telecom Video: http://www.opencem.org/cimi
• Co-Chairs: Stan Huff, Linda Bird, Galen Mulrooney, Richard Esmond; where, quorum requires 2 co-chairs
• REQUESTED ACTION: Directly edit this WIKI page or send your feedback to CIMI@lists.HL7.org with your comments, questions, suggested updates.

Minutes (Annotated Agenda)

• Scribe: Richard Esmond
• Telecom Audio: https://1drv.ms/u/s!AlkpZJej6nh_k7906pJl4wSDGh7sMg
• REQUESTED ACTION: Update wiki directly or send suggested changes to Scribe or cimi@lists.hl7.org
• Bolded Items were discussed/annotated to the agenda

• Record this call
• Agenda review
• Introduce new participants
  • Yvette Apura, OD, RHIA (from PCPI)
• Meeting schedule
  • Stan will be gone the next 4 weeks
• Review and approval of minutes
  • http://wiki.hl7.org/index.php?title=CIMI_Minutes#CIMI_Minutes
  • Approval of minutes: Richard/Susan; 17,0,0
• Meeting days for San Diego WGM – Galen
  • We have reserved rooms to meet Sunday thru Thursday, either alone or jointly
  • September Tooling pre-meeting in San Diego
    • Tooling session hosted by Claude starting Wednesday
    • Possible Friday cram-session
    • Richard will send a note in coordination with Claude
    • We can use Sunday for cram-session as well
• Brief updates on active projects (standing item)
  • Brief update on VA KNART effort – Claude
  • NIB for September – done!
    • CIMI will ballot the CIMI Logical Models in September as Informative, move the ballot from the current Patient Care PSS to be under IIM&T PSS, or other solution work with HL7 HQ to move the ballot to a CIMI PSS.
    • (Galen/Jay) passed: 14-0-0
  • Consider using Jira for ballot-comment tracking
    • Become part of the HL7 pilot of Jira as soon as possible
  • Skin and wound assessments – Jay and Susan
    • Representation of cross-hairs on an image for location of wounds or other abnormalities.
      • We will bring this back next week
      • Get people from DICOM to share their knowledge
      • Ask FHIR people whether they have dealt with this
  • Regeneration of lab models – Susan, Patrick, Joey, Stan, Galen, Richard
    • 2nd meeting happened on Tuesday May 30
    • JP-Systems may have a resource to contribute
    • UCUM and SNOMED CT Units of Measure – Susan Matney
      • Everyone to send any cross walk that they have to Richard
  • Conversion of CIMI archetypes to FHIR logical models to Profiles – Claude
  • FHIR resource profile from FHIR logical profiles –
    • Claude and Richard are collaborating using Travis from PenRad as additional resource
  • Creating ADL models from CEMs – Joey
  • LOKI – Patrick Langford
  • CIMI Website – Patrick Langford
  • Tool that takes the BMM patterns and produces FHIR profiles [Richard, Michael van der Zel]
  • MDMI tools – Steve and Richard to follow up
  • Help to create documentation tool that reads a model and generates wiki pages (a document generation tool) [Claude, Michael van der Zel]
  • Review Argonauts profiles and FHIM classes for vital signs – Susan, Galen, All
  • FHIM – CIMI integration – Galen
  • Cancer DTR (Diagnosis, Treatment, Research) project – Richard
• Report of CIMI webinar to FHIR Infrastructure - Claude
• Unification of the EvaluationResult and Assertion models – Claude and Joey
• Someone needs to send Claude the latest ballot comment spreadsheet so he can propose dispositions for modeling comments and see who we can recruit at HL7 to help us with specific models.
• What is our approach to modeling of procedures, surgical as well as imaging, etc.
  • What are the rules of what to pre and post coordinate?
  • We will bring this back to the agenda when we have more information
• Review outline of Jay’s SCT paper
• Proposals for additional principles:
  • We want to do this when Jay and Stan and any other interested parties are all available
    • Richard will check with Jay
  • The proof of the architecture is in the results. Architects may apply principles as needed, but if the architecture has to contain inconsistencies in order to deliver clinically correct archetypes, that is appropriate.
    • Corollary: Use the architecture to make models and test the architecture by making the content that we need. The architecture does not need to be perfect before we make models and try them out in implementations.
  • It is strategically important to assign stable and unambiguous concept identifiers to coded elements in CIMI resources. It will also be important to ensure that these concepts are modeled appropriately to support logical classification, but this requirement can be deferred, and should not hold up CIMI development.
    • Models and model elements will have unique identifiers that can be used for creating relationships in an ontology. However, the complete positioning of the models in the ontology should not hold up CIMI development.
  • While CIMI does not need to ensure the correctness of the SNOMED CT graph, we do need to lay the foundations to support accurate logical classification in the future. [This general principle should form the requirement underpinning specific proposals for semantic model alignment, e.g. Evaluation Result.]
  • It is okay to declare an attribute in the parent class and refine its semantics in downstream classes if the semantic restrictions represent subsets of the ancestor attribute's domain. The children should be created by restrictions on the range and domain of attributes in the parent. Attributes that are specific to the child only would just be added in the child. (Example: specializations of devices.)
    • We previously agreed that the semantic binding of the attribute (i.e. device, or body location) would be carried from parent to child, but the attributes could be renamed using the binding capabilities of ADL. Tooling can leverage the common binding to know that the name in the child is related to the name in the parent to know that “dispense device” is a device.
  • We make the assumption (requirement?) that implementation platforms have the responsibility to provide a direct path for retrieval of IndividualClinicalStatements regardless of what panel or panels contain them. The logical model provides enough information to support this behavior in the implementation. (The logical model can be different from the implementation.)
• Continue ballot reconciliation – All
• Review of updated assertion/evaluation table content – Stan
• Graph/STAMP modeling paradigm – Richard
  • We will do this every other week in place of the “Patterns” meeting
  • People can review and edit the existing description and requirements at:
    • https://docs.google.com/document/d/1VuEUXNHe4IE0DsukbU0PFDEJC4YYFYLmueqwC78s7G4/edit
• Planned work for September ballot
  • Models for lab data and regeneration of leaf node models (Stan, Joey, Galen Patrick, Susan, Katy Holck (pub health, lab), Donna Redley RN informaticist). See 2015 models.opencimi.org as baseline
    • Work needs to be done to improve the quality (value sets)
    • Include validation (compare CEM-LOINC (axes) structures to CIMI & FHIR)
      • e.g., fetus.heart
    • Start with highest volume and most used data elements
    • Use the SNOMED description logic maps that were generated as part of the SNOMED – LOINC agreement
    • Use of the “invariant” anchor patterns, Grahame’s “dictionaries”, abilities to do transformations of instance data
    • Claude – hierarchy of types: quantitative, coded, ordinal lab, detailed clinical models
  • Vital Signs
    • CIMI models for vitals map-and-gap against FHIR models
    • Compare/validate to/with US Core/FHIR Core, Intermountain CEMs, MHS Cerner, VA
  • Document refset requirements (intensional, extensional) for CIMI binding and tooling
    • STAMP versioning, implications of versioning and model dependencies for concepts that are referenced in CIMI models – Susan Matney
  • Clarification of binding – static versus dynamic binding, and related issues, allows conformance testing, binding in abstract types – Rob McClure
  • Further flesh out core reference models – Claude
    • Allergies/Intolerance/Adverse events (Claude working with Russ)
    • Action (flesh out stub that is in the May ballot)
    • Event (flesh out stub that is in the May ballot)
    • Medications: order, administration, dispense (FHIM)
    • Devices (FHIM model very mature IAW V3 & FHIR, pull in device WG)
      • Claude: governance issue
      • Wait on devices till medications done
    • Subtyping of results for radiology and imaging
    • Care plans (Richard: CDS & CQI working on this)
      • Susan & Stan: Multiple patterns (panel on observations)
    • Harmonizing with QDM, FHIM, FHIR, CQI, VMR
  • Tooling (Claude)
    • Galen: CIMI do reference-archetype models, in BMM, and crowd-source foundational patterns & DCMs
      • BMM level 1-3, Data types, foundational models, (FHIM) clinical architypes
      • BMM “constraint” level 4: Patterns & semantic anchors (Who is responsible? (CIMI vs crowd-source)
      • BMM “constraint” level 5: DCMs
    • Generation of FHIR profiles from CIMI models
    • Model authoring tools
  • Generate BMM from harmonized FHIM, (Galen harmonized FHIM target date July 1)
  • Model request spreadsheet import (Susan & Richard)
  • Strategy for testing of models (Stan)
  • Pharm (Claude): CIMI vs. pharmacy models. Claude & Galen meet during Pharm meetings to align patterns and FHIR resources and patterns. Pharm be source of requirements and own the harmonized logical models.
  • 2018 Option (Richard): Extend process to radiology
    • Radlex orders and results/findings separate; where, findings are lower quality
    • NANCY: shared procedure file in 2018-2019.
    • Map LOINC to CIMI DCMs
    • Tooling, databases (SNOMED & versioning) and process guide for scaling concept creation, distribution, CIMI binding-and-refsets (in SOLOR),
• Any other business
• Future topics
  • Loading of concepts into SOLOR – Susan Matney
  • The SOLOR Conundrum – Steve Hufnagel
  • Review CIMI Observation Result pattern - Stan
  • How will CIMI coordinate with DAF? - Claude
  • Granularity of models (schematic anchors) – from Richard
  • We need a way to identify the focal concept in indivisible and group statements
    • We would probably use the new metadata element
  • New principle: Don’t include static knowledge such as terminology classifications in the model: class of drug, invasiveness of procedure, etc.
  • Proposed policy that clusters are created in their own file – Joey, Stan
  • The role of openEHR-like templating in CIMI’s processes - Stan
  • IHTSDO work for binding SNOMED CT to FHIR resources – Linda, Harold
  • Which openEHR archetypes should we consider converting to CIMI models?
  • Transform of ICD-10 CM to CIMI models – Richard
  • Others?