HL7 Work Group Meeting - CGIT Notes

San Antonio Texas – January 2015

Monday, January 19, 2015

CGIT Q1: Administrivia

Agenda for meeting

  • approve agenda
  • Other than V2 conformance, joint meetings with FHIR on Wednesday afternoon, Frank takes lead on this. For CDA we don’t do much. Still maintain RCNL.
  • One of our projects is to create an overall standard of general conformance that all standards (V2, CDA, FHIR, etc) can use. Details will be different but basic concepts should be the same.
  • This afternoon we can write up the proposal.
  • Cancel Q4 for today.
  • Talking about Tuesday Q2 tomorrow
  • Think about creating presentation for Paris on Conformance about AIRA-NIST project
  • Nathan needs to reach out to Mario at Aegis and invite him to the meetings
  • NIST is currently leading the process for creating new proposals for how to write implementation guides, this process has precedence from Message Work Bench tool where changes were requested and later they were rolled into the standard for V2.
  • Talking about how Acknowledgements should be used. Talking about value sets and how they are bound to the field. The question is why isn’t this in a general document? We need to work on this because FHIR has no conformance and value set control. Implementation guides are very different between V2 and V3. Need to volunteer more aggressively with other groups.
  • Product Line Task Force, need to invite someone from there for Q2 tomorrow
  • Question about Paris. Expect a lot of first time attendees. How HL7 does conformance should be in one place. Lots of documentation is stale. It’s not good that there is splintering in how the work groups work on projects.
  • Need to plug PSS into the product line. People won’t come to us, we need to go to them. Iona will send PSS to Mary Kay, Austin, Tony Julian. CC Ken. PSS is called “Unified Conformance and Constraint Modeling”. We will need to refine with them. MnM, Vocabularly, and InM. Need to add Product Line Task Force.
  • There are issues with FHIR and conformance. Talking about extensions.
  • approve minutes (Chicago)
  • review and approve DMP
  • Voting on DMP, keeping the rules the same and changing the name. Nathan made motion, AM made second. 4 for, 0 against
  • agree on conf call schedule (currently Mondays 4pm ET)
  • review work group health metric (resolve issue if necessary)
  • conformance facilitator: role and responsibilities
  • follow-up action items
  • projects (review)

Q2 v2.9 Proposals

Agenda

  • Message Profile Identifiers (Managing multiple versions-OIDs)
  • Came up with solution in the last meeting, but nothing has been done
  • Managing Value Set OIDs (How to specify in MSH.21)
  • Rob Snelick has 60 slides in this area, haven’t formalized yet
  • Harmonized Data Type Flavors in Profiling (e.g., TS_1, TS_2, etc.)
  • Still proposals that need to be written
  • Requirements on data sets should not be embedded in data types
  • Will be pulling this out and making a separate document, not in the main guide, that way it is independent of the HL7 v2 release, could be applied going back to previous versions
  • Need to separate into specification for XML and document that explains it
  • Could use wiki for this, or HL7 GForge
  • Compiling a set of links for a developer
  • Need more resources in this area, not sure when this group will have the resources to finish this and bring it out. It’s been two years in discussions, a lot of progress at NIST, but no time dedicated to moving this out as public resources.
  • Frank Oemig can now generate Chapter 2C with all the code values from his database. This work is helping to define exactly what a value set is, code set is, a code table is, etc.
  • Specifying explicit conformance statements in the base standard and IGs
  • Replace CWE/CNE/etc. with simple and complex coded element??
Discussion on Paris Agenda

Want to do mini-tutorials. What do we want to discuss? Going through the agenda, a lot of discussion. Rob Snelick is writing up the proposed agenda. HL7’s role is documentation and education.

Q3 Paris Agenda

General discussion before the meeting: Talking about US versus International standards. Not many people even know about DSTU comments. Someone needs to be make a big deal about it so people can give feedback on standards. Simple website to channel comments. Members voting 15%, Germany 2-3% vote on localization guides. People are not familiar with the process.

Agenda looks good.

  • Monday Q3: Overview
  • Monday Q4: Testing tools

Talked about new book that Rob Snelick and Frank Oemig are working on. This book deals with conformance testing in general.

CoChair Meeting

Breaking for Foundation and Technology Steering Division (FTSD)
  • Roll Call
  • 30+ attendees
  • Approve minutes & accept agenda
  • Agenda accepted
  • Block approved
  • Room scheduling
  • If you are having joints the person setting up the joint needs to set it up or you will get blocked
  • Site has not be available until after the work group meeting, not needed for Paris.
  • Its working well but for a few things
  • Soliciting mentors for first time attendees
  • Nathan Bunker volunteered to help first time attendees
  • FHIR WG Proposal discussion
  • Proposing a new work group that will be tasked with maintaining a core set of elements, other workgroups would develop additional FHIR resources, etc.
  • Fire Management Group would continue its same tasks. It is the decision authority, but does not develop content, but work group has no decision authority.
  • Looking to eventually create a single governance group and this will oversee governance and content.
  • Discussing the concept of this new work group. The discussion will get tabled for another day. Need a diagram and text describing the bigger picture.
  • Administrative documents: SWOT, DMP
  • Elections
  • Some work groups have elections
  • Help Desk and Work Group Relationship
  • Volume of questions is very low, about 6 per month, only 4 of them made it to the work groups
  • If we get questions we should forward them to the helpdesk when they are resolved. That way the help desk is in the loop and can add the response to their list so they can help in the future.
  • Work Group Health / PBS metrics
  • Handing out stars to work groups that are in good health

Tuesday

Q1 Work Group Health and Tooling

Work Group Health

  • Harmonization, check the next cycle, we are not participating, send email to Ann or Karen. We need to check for changes to V3. Need to state that. Action item: get Harmonization list serve. TODO: Connect back with Frank on this in the next week
  • DMP: Should be a standing item for the first meeting year, need to state that in our minutes. Next FTSD call will tell them we reviewed our DMP. Should also need to do SWOT. Put reminder for DMP every year.
  • Project Insight, send email to Dave Hamill to get access. CGIT/CGIT or ITC/ITC is the password. Update the project due dates.

How to run meetings:

  • Review minutes from last meeting
  • Review agenda
  • Review projects

TODO Nathan Bunker will get access to Wiki.

Messaging Work Bench

It is ready for people to download. You have to install the original version and the download the patch. Updated December 2, 2015. TODO Rob Snelick needs upload HL7 v2 Conformance Profile schemas. They should go on the HL7/CGIT document server. Rob will also put it on the wiki too. Need to find from Matt Gram what version he is at. TODO Get updates from Matt Gram at regular intervals.TODO create a list for MWB.TODO Update web page linking to MWB list serve and to google groups for NIST tools and Aegis tools, and any other tooling.Discussing about functionality that is needed in MWB for saving and loading profiles.

NIST HL7 v2.x Message Validation Tool

Supports ELR, Vital Records Death Reporting, and Height and Weight. Discussion about problem with using templates or profiles to identify what information should be in a message. HL7 v3 can identify multiple templates throughout, but in V2 there is only one place to put profile identifiers in MHS-21. Talking about registering OIDs for profiles and EI data type. EI data type is badly designed and often misused.

Q2 RCnL Discussion

Refinement Constraints and Reconciliation. Ioana is sending a note about recirculating an RCnL affirmation.

TODONathan will take over the attendance list for the group.

TODO Nathan will create a work group help plan to ensure that everything is being done on time. From that he will create reminders to have things done in the right time.

Insight on FHIR. Ioana is going to look at code that Lloyd pointed out that validates a profile. Discussing how FHIR is doing conformance. Talking about how FHIR only focuses on 80% and thus will have to have lots of extensions to be used. We need a lot of resources to represent the content. Talking about the CDA to FHIR project. Currently the plan is to map the concepts over to FHIR Resources. Ioana is in favor of just moving the current CDA objects directly into FHIR and create new resources. Then clean up from there. With FHIR we lose modeling (RIM is gone) and then we lose support for conformance. Then we lose control with constraints. Ioana talked about two systems that worked to integrate and they evaluated every field, trying to map them to a common structure. They found that they successfully mapped the right field, to the right field in the other system 60% of the time, 40% of the time they made a mistake. Frank talks about the two things we must do: Documentation and Education. Local sites cannot interpret it locally, only the community can. Rob says we should have validation tools before anything can be balloted. Without validation established how can the standard be finished?

Wednesday

Q3 & Q4 FHIR, CGIT and Templates

CGIT comment (from Frank Oemig): We would like to discuss how profiles are created and maintained. How we can ensure conformance compliance. Fear with FHIR is that a huge set of profiles are being invented, how to ensure consistency across them. Should be a clear mechanism for how it is communicated.

FHIR reply: Profiles can be chained together.

CGIT: What are the rules for adding extensions and constraining?

FHIR: Currently implicit rules but not explicitly stated. Currently missing from FHIR specification. At this moment anything can be done. Even data types can be changed. Clearly we would like a set of exact rules so tooling can implement them. Cannot constrain extensions, only constrain the use of the extension.

FHIR is walking us through DSTU 2. Currently called “Profile” will be changing. Explaining how the model works and how to get to it on the website. There is a section on Conformance. It would be great help if we from HL7 provide rules that are reasonable when making constraints. Discussing backwards/forwards compatibility. We need to put this all on rules on paper, this is what is missing from the specification. How are we going to work on this? Idea to work on this in Q4 to make a bullet list point to start the ideas for conformance.

Reading email from Lloyd McKenzie and talking about creating a conformance document. Group talks about general document about conformance. Need a general principles of conformance document.

Question about changing elements. Annotations can add clarity, but don’t change the elements.

Breaking for 30 minutes

Will be discussing Constraint Types. Kai is starting from the RCnL document and pasting into the new document. A start document is created. Kai will do some adjustments and then send it out. We missed a lot of rules about slicing

  • Slicing: 0..* references to observation, but want to slice that into 1..10 but 3 must be observations of subclasses of this LOINC code and 5 must be subclasses of LOINC… Need to align slicing with how it can be done in HL7 Version 3. HL7 v2 has something like this as well. In principle is more-or-less the same. Similar to concept of co-occurrence.
  • Open EHR has a set of rules similar to what we have been writing. Would like to check that document to see if we can reuse parts of this.
  • Would be good to explicitly number the rules so the tool can reference the rule. Seems to be a good idea.
  • Discussing meaningWhenMissing and defaultValue.
  • Missing discussion about extensions about excluding extensions. Same constraint rules should probably apply.
  • Operation definitions can be based on other operation definitions. Need to look at operations and how they can be derived. Need to support some changes (such as different address types/currency) but not to allow new parameters that might change the operation.
  • We have some topics for vocab because not everything is fleshed out in RCnL. Vocabularly binding concepts are so complicated. It will make things easiest if we re-use concepts around vocabulary that we have already developed.
  • Profiling FHIR

Question about how much of the standard that vendors have actually implemented. How far is this done according to the specification? Comment from implementer that they didn’t look at conformance to profiles as this was expected to change in DSTU version 2.

Now Looking at FHIR ballot items:

  • Writing a set of constraints against a profile. Sometimes you just want to append something to a paragraph. Group in favor of allowing … to be added to start of text, to expand previous text in the new description. Ewouk proposed, Nathan Bunker seconded. 12 for, 0 against, 1 abstention,

Motion to adjourn.

Thursday

Q1 Vocabulary

Talking about Vocabulary Binding syntax and PHIN-VADS. Talking about work to move codes out of PHIN-VADS and put in VSAC. PHIN-VADS will then be focused on just public health items and so should stay up-to-date in these areas. All other codes will go to VSAC.

Binding Syntax

Need to get Binding Syntax project back started. Going over previous events:Decision was made to start a project to create a human readable version for implementation guide. (MIF is hard to read.) Got syntax into BNF. But structured documents doesn’t follow core principles. Long discussion about should vs shall. Stalled since last summer. Do we still have the appropriate goal? If not then change it. Should we cancel the project? Who is going to actively work on it? Rob McClure will take the lead on the project and work on the document. Wendy Huang will help too.

External Chapter for HL7 version 2

Vocabulary and conformance need to be broken out of the standard and tracked separately. Action Item: need to take item up to TSC.

Q2 Vocabulary

V2 Code Tables Project

PSS is here:

Ted is giving background on the project. Open question for discussion: Bindings for external tables that we only use a subset. Examples or representative? Ted wants to go through the 20 something tables that might be representative.

Second issue: request from CGIT for a paragraph for 2.9. Will go in Chapter 2C to be a message to the implementer. And it will be in conformance section as well. Need to put a link in Chapter 2C to go to the conformance chapter. Rob Snelick: how to profile and constrain the vocabulary. Educate implementation guide authors to correctly specify and define vocabulary in V2.

Third issue: Ted would like to see a recommendation that folks use the latest published vocabulary regardless of the standard, except the structural part of the standard. Making a request to CGIT to add this to their conformance standard. CGIT will add this. TODO: Add action that CGIT is going to do.

Fourth issue: Discussing issue with odd code tables. Values that are unbounded, and ones that are made of pieces.