20100617 arb minutes

From HL7Wiki
Jump to navigation Jump to search

Template:Arb minutes Architecture Board June 17, 2010 Late Call


Attendance

Name PresentAffiliationE-mail address
Bond,Andy Yes NEHTAandy.bond@nehta.gov.au
Curry, Jane Yes Health Information Strategiesjanecurry@healthinfostrategies.com
Grieve, Grahame No Kestral Computinggrahame@kestral.com.au
Julian, Tony Yes Mayo Clinicajulian@mayo.edu
Koisch, John Yes Guidewire Architecture jkoisch@guidewirearchitecture.com
Loyd, Patrick No Gordon point Informatics LTD. patrick.loyd@gpinformatics.com
Lynch, Cecil Yes ontoreason LLCclynch@ontoreason.com
Mead, Charlie No National Cancer Institutemeadch@mail.nih.gov
Nelson, Dale Yes II4SMdale@zed-logic.com
Ocasio, Wendell No Agilex Technologieswendell.ocasio@agilex.com
Parker, Ron Yes CA Infowayrparker@eastlink.ca
Quinn, John No Health Level Seven, Inc.jquinn@HL7.org
Shakir, Abdul-Malik No Shakir ConsultingAshikir@COH.org
Guests
Haddorff, Richard Yes Mayo Clinichaddorff.richard@mayo.edu
Hufnagel, Steve Yes U.S. Department of Defense, Military Health System Stephen.Hufnagel.ctr@tma.osd.mil
Koehn, Marc Yes Gordon point Informatics LTDmarc.koehn@gpinformatics.com
Laakso, Lynn Yes Health Level Seven Internationallynn@hl7.org
McGaughey, Skip No
Peres, Greg No CA Infowaygperes@infoway-inforoute.ca
Robertson, Scott Yes Kaiser Permanente</td?scott.robertson@kp.org
Rospide, Eddy No Albany Medical Center rospide@mail.amc.edu
Smith, Karen Yes HL7 Technical editorkaren@smithspot.com
Thompson, Cliff Yes OntoSolutions LLCcliff@ontosolutions.com
Thompson, Cliff Yes OntoSolutions LLCcliff@ontosolutions.com
Wrightson, Ann No HL7 UKann.wrightson@wales.nhs.uk


Agenda

  • Call to order
  • Approval of Agenda
  • approval of Minutes

June 10, 2010 Minutes

  • Peer Review - Need help with reconcillation.
  • Discussing the IF from Cecil
  • Harmonization of Project scope statement
    • HL7 Harmonization Framework and Exchange Architecture
    • (HF&EA)
  • Touchpoint on IG.
  • collaborate with SOA on managing expectations for CTS2
  • OHT pilot value proposition survey
  • Other business and planning for next meeting
  • Adjournment

Call to order

The meeting was called to order at 6:00pm U.S. EDT with Ron Parker as chair and Tony Julian as scribe. Quorum was achieved.

Approval of Agenda

The agenda was approved by affirmation

approval of Minutes

MMS Jane/Tony to approve June 10, 2010 Minutes (5-0-2)

Peer Review - Need help with reconcillation.

Ron Parker - looking for volunteers to help with the triage. If we have people who can give us an hour. Tony will assign names, with a slot, and consolidate. Tony and I dont have enough cycles. Jane, Steve, Andy, Tony, Cliff, Ron.

Discussing the IF from Cecil

Cecil Lynch: I hae defined and dirculated an introductory part that referes to an example with a DAM going down to the attribute level to explain why we need eccf. Explain why a middle-in approach leaves you gaps. A full v3 class structure cant be expressed in 11179. I am done with DAM, the difficult part, there is no concensus on what a DAM is except for an introduction from Rene. Taking information from e-mail threads, with v3 guide. Passing by Lloyd at each step. have not been able to work on it this week due to deadlines. Hope to work on it this weekend. Karen has my concept map, I will be sending her the elements in order, and will be able to wrap up next week.

Jane Curry: DAM Have you run into the issue where the same domain is viewed by different groups with different understanding of the terms?

Cecil Lynch: Yes - bridge model has 5 views of the model. While the classes are re-used, they dont have different interpretation. There is the opportunity to interpret it differently. I have comments on the HF&EA. In HL7 we have definition inconsistancy. There will be a lot of comment on that - I cant write something everyone will agree to. DAM's have different levels of abstraction. And end user business process model that can be interpreted by a person who can build HL7 model is not accomplishable. May be an issue with my cross-training in medicine - expressing understanding by physician as well as a modeller. Will need comment by reviewers at another level of understanding.

Steve Hufnagel: Do you have an early draft that we can comment on?

Cecil Lynch: Hope to get a copy to Karen tonight, but it may be this weekend. I still have writing to do.

Steve Hufnagle: We are doing the computationaly independent model for EHR.

Cecil Lynch: I have some comments, and disagree on the CIM contents.

Ron Parker: Thanks Cecil. We are hoping to see the draft. There are some things that the IF will shape and refine, and answer questions.

Cecil Lynch: Will provide forum for understanding.

Ron Parker: Jane and I have had two calls on governance, which will change the way we do the BF. Each piece informs the others. It will be helpful. Even if we can get the drafts out, it will help us consolidate our opinons about the disposition on the peer review.

Cecil Lynch: If the team would like, i could send out each section as finished if that would be helpful.

Steve Hufnagel: That would be good.

Ron Parker: We will give Cecil some latitude. It would be Great.

Harmonization of Project scope statement

HL7 Harmonization Framework and Exchange Architecture

Steve Hufnagel:After doing the Immunization case study we found granularity issues as well as definition issues. This fostered this proposal for a SAIF implementation guide, with a set of appendixes, to be familiar with various methodologies. Ours would show how DODAF/MODAF would fit. Canada would have a TOGAF version, and a Zackman appendix. There is a desire to see a agile scrum appendix.

Jane Curry: Have you talked to Berndt? he has a framework for distributed components. Steve Hufnagel: If you are willing to spend the sweat, we can have an appendix for your favorite flavor. We will expand the appendixes by stakeholder participation. Any of them could be an IG using the appendices. You dont have to use all of the artifacts for every project - notional set, cannonical set - using UML. Will have topic map showing the relationship of the artifacts. The BF ties the pieces together. Any artifact in the BF may point to other artifacts in the viewpoints. Goidn theought the HL7 development framework you can use the components to define business objects, to define services/messages, and therefore constraints, documents, messages, and services. A fully qualified ECCF is a comprehensive Interoperability specification. It is important to show that process to build a viewpoint into a test or aquisition specification.

Steve Hufnagel:We are proposing based on HITSP experience, a small number of people do a first draft prototype - drafted a TOC from TN904 from the HITSP document. Chapter 3 is conceptal/management view, Chapter 4 is technical view. The harmonization framework Model is a conceptual map. Chapter 5 HL7 Exchange Architecture defines doing object generation from the HL7 artifacts and the information exchange model. Chapter 6 defines standards and their categorization. Chapter 7 defines use declarations within a selected standard.

Steve Hufnagel:there is a list of concepts - the most important is the difference between information models and DAMS. The work that Cecil has been taling about identify verification/validation checkpoints(DAM, CDA, CMET, D-MIM). After you finish the analysis you go to the design to create DSTU /Interoperability Specifications. This is a model driven approach, so you can generate from the models. There would be a checkpoint, then you would have a SAIF ECCF interoperability specification. Analysis defins DIM, Design the PIM, PSM at DSTU stage, ECCF at implementation and test.

Steve Hufnagel:The framework I presented in Brazil was the results of the immunization project shows the artifacts that can be re-used at each place in the ECCF Specification Stack. Defining DMIMs you would be taking advantage of the other HL7 artifacts. The framework defines everything but the WHO - which needs a RASCI chart for each viewpoint - defining stakeholders and responsibilities.

Steve Hufnagel: First step is to develope a communication plan, socialize it, get buyin before going forward. Then do prototype based on HITSP work, the 80% that is defined. Then work through the appendices as validation. Doing DODAF we found it has a complete set of services, that fit in the engineering view. The service views that are similiar to the DODAF system views, fit into the engineering viewpoint.

Cecil Lynch: How did you decide to procede with topic map instead of concept map first? Steve Hufnagel: We needed to show dependancies - since Charlie Mead was doing topic maps, we thought to parallel that. We could do a full-up UML model, but it may add complexity, and make it less usefull.

Cecil Lunch: I would do the concept map using the ?? tools, so you can export in XTM and OWL. There is a statement about "However, in a data model, a single piece of data should be identified only once and be associated with the specific subject it describes. " What are you getting at?

Steve Hufnagel: It was originally the EHR data map. HL7 is vague with lots of use of 'information' in each. I used the Canadian definitions as a starting point. These words mean what we say they mean. You wind up with long philosophical discussions from various camps. I drew a set of definitions from Canada as well as researching in Google. I was starging with HITSP and IHE definitions. I dont have the viewpoint into all of the HL7 workgroups.

Cecil Lynch: "A Data Model does not identify or define any topics that are essentially derived by any transformation process." is a problem. Steve Hufnagel: There design is for the data modules as a set, which you use to bind in the value sets.

Cecil Lynch: We have a derivation component. It allows a place to communicate those topics. You can derive an expression at that point.

Steve Hufnagel: Certain classes need definition at the domain.

Ron Parker: In an EHR world where you are blending data from various viewpoints, the data model is fine grained, but built in a way that is compositional. Steve you are correct about the Canadian Approach. We build a granular model. When people look at a fine grained model, people dont see information, their own stuff. Use the same model, putting the informational viewpoint around it. When you reframe it in a messaging model, this is what you want.

Cecil Lynch: If I am going to build a data model in EA for generation of my database . . .

Steve Hufnagel: In the EHR project we got away from the Data model term, people were alarmed.

Cecli Lynch: You are using data model. . .

Steve Hufnagel: It is overloaded

Cecil Lynch: Defining a class you are defining a data model. If we consider - is a CMET a data model or information.

Jane Curry: Data

John Koisch: Informaion

Cecil Lynch: it is a framework for your information, that you can derive from.

John Koisch: Some is platform specific -provides backing.

Cecil Lynch: It is hard to get to a pure HL7 data model. We only have that at the core rim classes.

Ron parker: Steve - you woild like to take to TSC?

Steve Hufnagel: Yes. Ron Parker: It allows for one spec to be consumed by many. You need more time for realization - getting concensus on definitions will take time. Any other points before taking to TSC? ANy confustion?

Marc Koehn: THis would move the implementation forward. I need to know next converstaion.

Steve Hufnagel: Next step is communication plan.

Jane Curry: Pilot? Steve Hufnagel: Yes. To start socializing it. To let other provide feedback.

Ron parker: Any other comments.

MMS to approve the project scope statement subject to corrections offered by Cecil.  Tony/Cecil  (6-0-0) (only ArB members are counted in a formal vote).

Steve Hufnagel (21-Jun-2010): The Project scope statement was updated to include Cecil's comments and other suggestions. The Current HF&EA Project Scope Statement, information, documents & status are available at: http://hssp.wikispaces.com/HF&EA


Adjournment

the meeting was adjourned at 7:00pm U.S. EDT Tony Julian 15:50, 18 June 2010 (UTC)