MnM Minutes WGM 200705

HL7 Working Group Meeting
Cologne, Germany
April 29 - May 4

Sunday Q3 - Topic Review / Hot Topic Triage

Brief Agenda	Chair	Scribe	Room
Topic Review / Hot Topic Triage	Lloyd	Craig	Salon 5 - Bergisches Land

Hot Topic Triage (see MnM Hot Topics)

• Adding Record Target to ControlAct – May be appropriate for Friday
• Communication Process Model – Thursday Q2 (OO will be discussing this Tuesday Q2)
• Constraints on infrastructureRoot attributes - ?
• Context Conduction
• Domain Message Information Model
• Implementation of updateMode - To be closed.
• Model support for by reference
• NullFlavor
• Observation grab bags
• Packaging of Vocabulary with Static Models - Not believed to be an MnM issue.
• Participation
• Query Recursion - See action item below
• RIM Stewardship and Harmonization Representation
• Serialisation Annotations - Waiting for input from the ITS TC
• Serialization - Appears to just be Lore seeking approval - MnM will follow up on with this on a conference call
• TemplateId
• Use of IDENT Role Class

Action
Assignee	Woody
Item	Make a bug report or a tooling requirement from the Query Recursion Hot Topic.

Scheduling

• Monday Q1
  • Context Conduction
• Monday Q2 - Joint with SD
  • inconsistencies in application of methodology in CDA-derived specs
  • approaches to templates
  • model sharing between messages and documents
• Monday Q3 - Joint with INM
  • Dynamic assumptions by INM as prep for Wrapppers 2
  • Function of CACT
  • Dynamic model - trigger event vs interactionId (Mark T)
  • Action 2024: Transmission Addressing (time permitting)
• Monday Q4 - cancelled
• Tuesday Q1 - cancelled
• Tuesday Q2 - cancelled
• Tuesday Q4 - cancelled (Vocab MIF session in Vocab TC)
• Wednesday Q3 - Joint with Templates
  • Constraint Language strategy
  • Conformance constraints in ballot
  • Incomplete models
  • Entry points
  • Other ballot issues
• Wednesday Q4 - Joint with the Project Lifecycle Team
  • Application of project structure to harmonization, hot topics, etc.
• Thursday Q3
  • Constraints on infrastructureRoot attributes
  • Model support for by reference
• Thursday Q4
  • Domain Message Information Model
  • Observation grab bags
• Friday Q2
  • Adding Record Target to ControlAct

Action
Assignee	Woody
Item	Will see if Ken McCaslin is available for the Wednesday Q4 discussion.

Sunday Q4 - No Meeting

Brief Agenda	Chair	Scribe	Room
Tooling update for facilitators	Woody	Craig	Salon 5 - Bergisches Land

No formal meeting

• Sundry tooling issues were briefly discussed.
• Russ demonstrated the vocabulary harmonization tool he has been working on.

Monday Q1 - Hot Topics

Brief Agenda	Chair	Scribe	Room
Hot Topics Context Conduction	Lloyd	Craig	Salon 5 - Bergisches Land

See Context Conduction

Lloyd gave an overview of the Context Conduction Hot Topic.

Context Conduction does not propagate across Roles. Grahame pointed this out as something that is true, but is not documented anywhere. We discussed the need to create a document that describes how Context Conduction, Negation Indicator, Inversion Indicator, and such work.

Grahame expressed concern that even if we were to document context conduction well, it is so complicated that people will still either not use it, or will misuse it.

The current situation is a result of implementing context conduction in a way that allowed for existing semantics to continue, namely, that if context conduction is not specified, the result is indeterminate and left up to business agreement between the sender and receiver.

There was discussion about whether we need a formal notation or just more explanatory text.

Kathleen expressed the need to have information in one CMET preferentially override the same information provided by another CMET.

Action
Assignee(s)	Woody, Mark T., Grahame, Kathleen
Item	We will define the requirements for context conduction and then update the documentation of context conduction on the wiki to describe the current way this is supposed to work. We will compare the requirements to the current state to determine a course for the future.

The Context Conduction Hot Topic has been updated with issues raised during this session.

Monday Q2 - Joint with SD

Brief Agenda	Chair	Scribe	Room
Joint with Structured Documents Inconsistencies in application of methodology in CDA-derived specs Approaches to templates Model sharing between messages and documents How can we make sure the information being sent in messages and documents are the same? How can resolve model discrepancies across committees?	Lloyd	Craig	Salon 5 - Bergisches Land

Bob Dolin expressed the need for more consistent committee content in order to enable the content of documents and messages to be consistent.

General Model Consistency

Bob Dolin expressed that many inconsistencies are likely to be due to lack of communication between committees. He suggested that if committees strived for more consistency with the Clinical Statement pattern we could improve the consistency, especially for constructs that do not naturally belong to one committee.

Keith Boone suggested that at least one member from each committee should be looking at artifacts from other committees. Tooling that check for proper derivations may be helpful, but are not sufficient.

Woody Beeler pointed out that some inconsistencies are due to artifacts created prior to the formation of the Clinical Statement project. In addition just because two things appear to be similar, it does not necessarily follow that they should be identical. Some of the differences may be appropriate.

There was discussion about how CMETs should contribute to consistency. Keith suggested that some content may be most appropriately managed as CMETs owned by some "Infrastructure" domain. He also stated that expanding Clinical Statements to handle everything is probably not the right approach.

The scope of the RIM and the scope of Clinical Statements are different. The RIM is intended to support everything that HL7 needs to do. Clinical Statments are more narrow in focus.

Liora A. suggested that we may need an "Administrative Statement" pattern to handle the things that are out of scope for Clinical Statements.

Bob Dolin stated that in CDA R3 the intent is to synchronize with the most current version of the Clinical Statement pattern.

Woody suggested that if we have a Clinical Statement and an Administrative Statement, we should have a higher level Infrastructure Model too.

Woody suggested that for the CDA to work well, it should be able to reference the most current version of models from other committees, rather than having to copy all of the content. Bob, Keith and Calvin B. had concerns about CDA versioning too frequently.

Tools that can validate model derivation (in particular derivation from something like the Clinical Statement pattern) would be a very useful first step.

Grahame suggested that either the Clinical Statement pattern is incomplete or that there should be one super-pattern.

Summary: Clinical Statements and similar constructs are useful in ensuring consistency. Derivation validation tooling would be very helpful. We need methods for things like Clinical Statements and things such as Administrative Statements to reference each other.

There was apparent consensus that Clinical Statements need exit points.

Motion
Motion	MnM and SD believe that the Clinical Statement concept has been productive and should be extended with exit functionality. We need similar patterns such as Administrative Statements. We may also need enhancements to CMETs for handling finer grained items. MnM and SD will forward the request to the T3F to promote the use of a statment/pattern approach find a home for these activities.
Result	Woody / Bob (30:0:5)

Templates

There was discussion principally by Lloyd, Grahame, Galen and Keith about whether or not Templates are static models.

Bob Dolin pointed out that the requirement to use the MIF to create templates is problematic because we currently can't express everything that we want to templates for using MIF.

The question was asked "do templates have to be computable?"

Lloyd asserted that templates must be:

• ITS independent
• Enforcable independent of human intervention
• RIM based

There was discussion about Lloyd's claim that templates must be ITS independent. Without a defined formalism, it is problematic to specify ITS independent templtes that do not require human intervention.

Those doing CDA implementation guides are pushing forward with "templates" for what their needs without an official HL7 fromalism.

This discussion will be carried over into the Templates discussion Wednesday Q3.

Monday Q3 - Joint with INM

Brief Agenda	Chair	Scribe	Room
Joint with INM (co-chair election) Dynamic assumptions made by INM as preparation for Wrappers 2 Function of CACT Dynamic model - trigger event vs interactionID (Mark T) Action 2024: Transmission Addressing (time permitting)	Woody	Dale	Salon 5 - Bergisches Land

Monday Q4 - No Meeting

Brief Agenda	Chair	Scribe	Room
Cancelled

Tuesday Q1 - No Meeting

Brief Agenda	Chair	Scribe	Room
Cancelled

Tuesday Q2 - No Meeting

Brief Agenda	Chair	Scribe	Room
Cancelled

Tuesday Q3 - Joint with Conformance

Brief Agenda	Chair	Scribe	Room
Joint with Conformance ballot reconciliation?	Woody	Craig	Salon 5 - Bergisches Land

Attendance:

• Woody Beeler (woody-at-beelers.com)
• Lee Coller (lee.coller-at-oracle.com)
• Craig Parker (craigparkermd-at-gmail.com)
• Richard Kavanagh (richard.kavanagh-at-nhs.net)
• Steve Wagner (steve.wagner-at-va.gov)

Reviewed comments from the Conformance ballot. See ballot spreadsheet for details.

Now that Conformance is a Technical Committee, we do not have a future need to meet jointly for ballot reconciliation. However we may want to meet jointly for other reasons. We will tentatively plan to NOT meet jointly at the September WGM.

Tuesday Q4 - No Meeting

Brief Agenda	Chair	Scribe	Room
Cancelled (Vocab MIF session in Vocab TC)			Salon 5 - Bergisches Land

Wednesday Q1 - Joint with Vocab

Brief Agenda	Chair	Scribe	Room
Joint with Vocabulary (x_Domains, Harmonization, Ballot presentation)	Lloyd	Dale	Salon 5 - Bergisches Land

Agenda:

1. Rules for approval process X-Domains
   1. Should vocab chair harmonization
2. Cleanup of Vocab parts of ballot
3. Review of VS binding document
   1. Design time binding vs. runtime binding
4. Motion to approve agenda SH/DN

Discussion:

1. Why approve rules in we are eliminating x_domains?
   1. They are effectively value sets bound at design time
   2. Binding requires realm approval. When we create an x_domain

When an attribute is added to the RIM, if cannot be added without a concept domain.

The constraint process to get to a message mnodel is not implemental until every codede attribute is associated with a vS that conatins all values that can be used.

This involves concept domains, subdomains, and evebntuall a VS, and at sending time, a member of the VS gets substituted in and sent. At any point in the constraint process a code can be asserted. The constraining can be performed statically or dynamically

At some time there is an assoiction between concept realm and value set. This is binding.

Some of the associations require different approval proceses depending on associate realm, direct association of coded attribyet

LM:

1. Domain and VS bound in this realm
2. This VS will be used in this domain in this attribute.

In harmonization we never approval value set/design tome bindings. They are part of the model process, in the balloting. A static model is tied to a namespace model.

Universal VS can always be bound. Non-universal There are atrs in the RIM that are bound to VS in the UV realm. Not realm specializable.

SH: When a static model is made and a VS specified, this is binding in the UV domain. LM: It is not in the UV domain in that everyone needs to use this static model, but whoever uses this must use this VS. Design time vs. runtime binding

RT binding has issues because it impacts every static model that can be created. If we are binding Lab Observation code concept domain to LOINC universally, all might not agree. If you say in a certain template we will bind a particular attribute to LOIND, i.e., no one will object.

TK: this needs to be written to the HDF.

Vocabulary Design Principals

1. Creation or modification of value sets
   1. Creation or modification of value sets that are bound as value sets to non-realm-specializable RIM attributes or datatype properties MUST be approved through an HL7.org approval process.
      1. Value sets bound to RIM attributes must be approved through harmonization
      2. Datatype property value sets be approved by a ballot process.
      3. Between ballots, changes to datatype property bound value sets must first be endorsed by InM and subsequently approved through the harmonization process. At the harmonization process, InM is considered the steward of datatype property bound value sets when evaluating proposal changes.
   2. Creation or modification of value sets design time bound to non-realm-specializable static model attributes or datatype properties MUST be approved through the harmonization process.
      1. Purpose of the approval is feedback and potential for re-use
      2. This approval generally qualifies as a “non-voting” harmonization item.
   3. Modification of value sets bound to realm-specializable attributes or datatype proprerties does not require approval at the HL7.org and usually does not require approval at the Affiliate level
2. Run-time bindingd between domains and value stes MUST be approved by HL7.org (Generic Binding Realms – Universal, Example, Representative and Unclassified) or by the HL7 affiliate responsible for the binding realm in which the binding occurs.
3. Design time bindings (static bindings) to domains, value sets or specific codes are approved via the approval process for the static models or datatype specifications in which the bindings are approved.
   1. When defining static models to be balloted in the UV namespace realm, attributes that are identified as realm-specializable in the RIM and abstract datatypes SHOULD NOT be design-time bound to values sets.
      1. i.e. we should not be constraining affiliates to only using codes from a specific code-system.
      2. Exceptions may be made for UV namespace templates where there is not an expectation for all implementers in all binding realms to make use of the template or as proper derivation thereof.

TK: So aren’t x_domains always statically defined? Yes.

Q2:

Motion: (TK/RH) Vocabulary agrees to take responsibility for chairing the vocabulary portion Of harmonization (12-0-0)

1. Tooling issues TK: There are many idiosyncratic problems with the RIM repository with respect to vocabulary.

Action: Process for cleanup and decision making for this needed to be written by vocabulary and vetted and approved by MnM. Vocab and MnM have to agree on the approval process. RH/TK/DN/WB/LM

RH: DO we want to extend ND ENHANCE EXISTING aCCESS database or go with a different model. We’ve already exported into Lexgrid; can use Lexgrid to extract and cleanup vocabulary {Woody, Ted, Russ, Lloyd and anyone else interested} to help do this.

TK: Does the existing tooling (Visio, RT, etc) run on top of Lexgrid? RH: No. TK: Tooling is migrating to the MIF. LM: We are now bypassing RoseTree. TK: Are there export capabilities? LM: No. TK: We then have disconnects in the tooling.

TK: Design-time bindings can not be managed with the existing tooling. Vocab needs to spend time with MnM to go over the bindings. LM: You shouldn’t have bindings that begin with underscore. TK: They are there. There must be a parallel cleanup effort from the ballot repository from Jan 2007 to deal with cleanup of stuff in Visio from MnM. Vocab can identify errors, but will not decide upon fixes.

2. Changes to MIF Bindings Document

Review of Bindings Document from Tuesday Q4

Action items: JC/TK Unlike runtime bindings, for design-time bindings, coded attributes and datatype properties SHALL NOT be bound to more than one value-set. (14-0-1)

4. Approval process for x_domains (continued from Q1)

Additional editing was performed on the Vocabulary Approval Principals.

Wednesday Q2 - Joint with Vocab

Brief Agenda	Chair	Scribe	Room
Joint with Vocabulary (x_Domains, Harmonization, Ballot presentation)	Lloyd	Dale	Salon 5 - Bergisches Land

Wednesday Q3 - Joint with Templates

Brief Agenda	Chair	Scribe	Room
Joint with Templates Constraint Language strategy Conformance constraints in ballot Incomplete models Entry Points Other ballot issues	Woody	Dale	Salon 5 - Bergisches Land

Wednesday Q4 - Joint with Project Lifecycle Team

Brief Agenda	Chair	Scribe	Room
Joint with the Project Lifecycle Team Application of project structure to harmonization, hot topics, etc.	Woody	Dale	Salon 5 - Bergisches Land

Thursday Q1 - Joint with HSSP

Brief Agenda	Chair	Scribe	Room
Joint with HSSP including harmonization of models with 'messaging' Ongoing discussion, review SOA4HL7 issues	Lloyd	Craig	Salon 5 - Bergisches Land

• Alan Honey gave an update on some of the activities of the HSSP, in particular a document that he and John Koish have been working on regarding Methodology Mapping. The document is still in progress, but is nearing maturity.
  • Discussed "Business Services" which are pieces of business functionality approximately at the level of "lab result" or "specimen".
  • Alan urged further evaluation of WS-CDL for use in HL7.
  • We discussed the concepts: "service", "interface", "contract", "message", "operation", and "interoperability paradigm specification".
  • Service conformance profiles consist of two parts: a semantic profile and a function profile.
• Woody pointed out the desirablility of having the ability to communicate the same kind of information via either a message or a service with minimal effort in supporting both.
• Woody suggested that we may want to extend some HDF editorial ability to the HSSP for those sections where there is commonality. We need to run this by Ioana. Lloyd is willing to coordinate with Alan on this.

Thursday Q2 - Hot Topics

Brief Agenda	Chair	Scribe	Room
Hot Topics Communication Process Model (Dynamic Model)	Woody	Craig	Salon 5 - Bergisches Land

Communication Process Model (Dynamic Model)

See Communication Process Model.

There is a real need for a project lead for the Dynamic Model project.

We discussed the possibility of holding a Dynamic Model meeting in conjunction with the June RIM Harmonization meeting.

Thursday Q3 - Hot Topics

Brief Agenda	Chair	Scribe	Room
Hot Topics Constraints on infrastructureRoot attributes Model support for by reference	Craig	Dale	Salon 5 - Bergisches Land

Thursday Q4 - Hot Topics

Brief Agenda	Chair	Scribe	Room
Hot Topics Domain Message Information Model Observation grab bag	Dale	Craig	Salon 5 - Bergisches Land

Thursday Evening - Facilitators' Roundtable

Brief Agenda	Chair	Scribe	Room
Facilitators' Roundtable	Woody	Craig	Saal 1 Maritim

Special Topics

• Discussion of an out-of-cycle meeting for Dynamic Model in conjunction with the June RIM Harmonization meeting.

Committee Reports

• Clinical Genomics (Amnon Shibo)

• Patient Administration (Norman Daoust)

• INM (Grahame Grieve)

• Vocabulary (Ted Klein)

• NLM Project (Ted Klein)

• OO, Lab, PHER (Patrick Lloyd)

• RCRIM (Mead Walker)

• Patient Safety (Mead Walker)

• Pharmacy (Hugh Glover)

• Structured Documents (Bob Dolin)

• Tooling (Woody Beeler)

• Clinical Decision Support (Craig Parker)

• MnM (Lloyd McKenzie)

Friday Q1 - MnM Wrap-up and Planning

Brief Agenda	Chair	Scribe	Room
MnM Wrap-up & planning for next harmonization & workgroup meetings Meeting Highlights Agenda for Next WGM Agenda for Harmonization Conference Call Schedule	Woody	Craig	Salon 5 - Bergisches Land

Friday Q2 - Hot Topics / Fast Track Issues

Brief Agenda	Chair	Scribe	Room
Hot topics/Fast Track issues Adding Record Target to ControlAct See Adding Record Target to ControlAct, Control Act	Lloyd	TBD	Salon 5 - Bergisches Land