20101104 arb telcon minutes
|HL7 ArB Work Group Meeting Minutes
Time: 4:00pm U.S. EDT
|Facilitator||Ron Parker||Note taker(s)||Tony Julian|
|.||Curry, Jane||Health Information Strategies|
|.||Grieve, Grahame||Kestral Computing|
|X||Hufnagel, Steve||U.S. Department of Defense, Military Health System|
|X||Julian, Tony||Mayo Clinic|
|X||Koisch, John||Guidewire Architecture|
|.||Loyd, Patrick||Gordon point Informatics LTD.|
|X||Lynch, Cecil||ontoreason LLC|
|.||Mead, Charlie||National Cancer Institute|
|X||Ocasio, Wendell||Agilex Technologies|
|X||Parker, Ron||CA Infoway|
|.||Quinn, John||Health Level Seven, Inc.|
|.||Shakir, Abdul-Malik||Shakir Consulting|
|X||Knapp, Paul||Continovation Services Inc|
|X||Laakso, Lynn||Health Level Seven International|
|X||Pech, Brian||Kaiser Permanente|
|X||Smith, Karen||HL7 Technical editor|
|Quorum Requirements Met: (YES)|
- Call to order
- Agenda review and approval
- Approve previous meeting minutes
Ron Parker: We had a conversation in Cambridge, with a subsequent conversation.
Paul Knapp: We had a project which was work that was being done at MITRE Corp. on exchanging content. Project was to utilize their work with Micro-its to come up with content model that people would be happy about. Out of that came the HDATA work - it is NOT an ITS. HDATA work came up with two items:
1. Framework for exchanging data with content at rest, e.g. EHR's. Format for taking messages/CDA/other content, like a zip, with elements and a manifest, and shipping from a-b. You can extract components out of a richer model, and just send the components. Includes a Restful API.
ITS is responsible for transports. Original hope was that we would get a definition of a restful transport, and a profile on how to use the transport for HDATA.
What you see is not a restful transport for HL7, just a restful API for formatted transaction.
ITS is concerned that typically ITS will develop something where ITS is not the domain. Data types abstract is MnM, not ITS. ITS is not a content domain, and felt not well suited as the lead committee on the project - did not have personnel to determine need. Can work on whether this meets those needs.
John Koisch: There were three issues:
1. Issue of project scope statement - one of the things it did was tied HDATA to micro-its - which is not moving forward. It was unclear that there were API and serialization model.
Paul Knapp: It was a research project - not disturbed that the output was not the anticipated outcome.
John Koisch: A governance thing - not necessarily ArB concern.
Ron Parker: From an architecture perspective, there are issues around how this folds into the whole fabric. ArB is interested - content belongs in MnM.
John Koisch: I do feel that ArB is concerned about where the content goes, and an architectural issue on the imposition of structure of balloted material, with RMIM's and DMIM's. HDATA breaks the structures. HL7 bounds materials by interaction types. HDATA breaks the information boundaries as well as the behavioral boundaries.
Ron Parker: That is the issue:
Wendel Ocasio: What does breaks down means?
John Koisch: What resources are described by RMIMS. It is one thing to say this is the information, another to expose a resource via an API where you can get/put all of those things.
Wendel Ocasio: restful concept.
John Koisch: No, the HDATA breaks information down differently than HL7 has before. I could have missed something that delineates the binding between HL7 normative standards, and the way HDATA will serialize standards.
Dale Nelson: I agree with a lot of what you said. We have two issues: 1. restful data API, as Paul outlined, we hoped would manifest itself to transport anything, and we would add implementation guide (part of the research). Part 2, finding that the HDATA record format - where is the home, expose a situation in HL7 where there was a desire to move HDATA from point a to point b, with no domain group as the initiator. We want to understand, maybe outside the bounds of the original project statement - needs to be incorporated into a domain committee. We have balloted. No one has spoken up, not good, not bad, but that is the situation.
Wendel Ocasio: HDATA will be balloted by HL7?
Dale Nelson: Yes.
Paul Knapp: HL7 is 3legged stool, messaging, CDA,SOA. HDATA may be a fourth leg. It is up to HL7 to decide.
Wendel Ocasio: It seems like it is a different concern. At a different level than other specs = it is just a container.
Paul Knapp: Talks about behavioral dynamics, but does not express them.
John Koisch: The dynamics of get-put are described by HDATA. This is the resource discussion. The semantics of update and put-ing data, HL7 has done well, but querying and the joins is another thing.
Wendel Ocasio: There is no dynamics of CDA in how you persist.
John Koisch: Except that they have not made that constraining.
Wendel Ocasio: It is so lightweight, it is just HTTP profile. It is nothing to get worked up about.
Dale Nelson: In between is the assumption is the effect of change in a data structure.
Wendel Ocasio: Rest is just about what a resource is and how it is CRUD. It is not a problem, it is not novel.
Dale Nelson: It allows full crud operations, it allows one to say that for the purposes of this EHR document this is how you add/delete/move sections. it belongs in domain committee expertise, including transactional control. We do not have it tied down by a committee who understands the transaction;
Wendel Ocasio: Do you want a full WS with control, or do you want to just have a transport. You have to choose simplistic or not. If so, do not use it. It does not discuss security, assertions, etc.
Dale Nelson: We are not disagreeing - it is enabling these behaviors, without the blessing of the domain committees, allowing it in their content. That is where we are concerned. Someone (ArB, TSC) should give us input whether it is right.
Paul Knapp: Just because we can do it does not mean we should.
Ron Parker: Mechanism in a lightweight fashion to deal with the resources. How would it be realized in current domain content? Does it diminish our capacity to understand context? Does is support our ability to define dynamic behaviors? It is not clear that rest is just another bundler of stuff. It can subset/take the place of the due-diligence of the standards we have published to date. It is an ArB concern, as well as others. ITS asked for help, and got no response.
I would like ArB to determine a constructive way to ensure that this is not a fourth leg as far a primary construction, make sure there is alignment, and the potential to use this.
Wendel Ocasio: Who has a business requirement? In HL7 we do not want to create a standard just for the exercise.
Dale Nelson: ITS is not the consumer. We shopped it to EHR. There response is we move CDA, it is immutable, so we do not want to blow it into little pieces to be re-assembled. The answer is that you could put entire CDA in HDATA, but that is wrappers on wrappers. It is a solution looking for a problem.
JK Project facilitator was Keith Boone - I would feel pretty happy if someone would say that HDATA would be good for packaging CDA, with HDATA as a serialization mechanism. Those are not the constraints I am seeing. There are lots of solutions looking for problems. When people look at an ITS, e.g. xml, they are looking at a paradigm for user. HL7 has for long time a massive incongruity of those who look at the RIM as a persistence model, others who see it for deriving messages. The Arb need to recognize that there is stuff this ballot/project could be constrained. RIMBAA was not in the best concern of HL7 - it could do anything. It is important for ArB to discuss how they fit together.
Ron Parker: I think that we need insight on the scope, what does realization of the capability mean? As an ITS you could implement any artifacts inside this construct, without rigor.
Paul Knapp: Without governance. HDATA is centered around content profile. HL7 needs to determine whether it enables something we want to enable, and meets our goals.
Ron Parker: If we need the dialog, without prejudging, we do not understand, to pop it out as a DSTU, is problematic - could create confusion. What process do we need to take it along on due-diligence? ITS could not get attention/energy. We need to provide advice to TSC as to their ability to control.
Wendel Ocasio: I do not know if there is time to affect the ballot. This lacks context, appropriate ballot needs context to other HL7 artifacts.
Paul Knapp: The organization should understand that it is preliminary. There is good effort in this - do not want to see good-will of the developers disturbed. Maybe DSTU is too high. Minimum back-off would be to determine whether it is a fourth leg, or needs completeness. Put it out for comment, as a minimum position? Nothing ever passes first go-around. We need in Sydney the attention to bring it into shape. It is process, but not construct.
Wendel Ocasio: Can we use SAIF to describe the constructs?
John Koisch: HDATA is really an implementation of a set of constraints on SAIF on how a SOA or REST is working.
Paul Knapp: SOA is seeing a use for this - they see a use.
John Koisch: That is the piece that speaks to me as being influenced by SAIF.
Ron Parker: There are several pieces:
Declarative statement about lack of context, not suitable for DSTU, it cannot be used as it. Organization does not understand. It is not ready for DSTU.
Paul Knapp: Things go through two ballots - if we change to for comment, then re-ballot in next cycle.
Ron Parker: Allows the due diligence. Got for comment, and to get to DSTU they need proper context.
John Koisch: In the SAIF, or IG?
Paul Knapp: In HL7 - we cannot determine if it is for us.
Ron Parker: When we say we need more context, what does that need to look like? Arb might take a run at it using SAIF concepts. We are not mature enough to mandate it.
Wendel Ocasio: We are not ready, but could recommend it.
Ron Parker: What are the constructs that are appropriate for this purpose? Do we have the energy and focus to take a run at it from a SAIF perspective? There are other dimensions - ability for the proponents to describe how domain models can be rendered? How to the resources correlate to cmets and other constructs.
John Koisch: We have restful implementations - using a constrained version of ARLIS at NCI. You will find that with other people. There is a lot of experience using REST - why you need an energized set of engineers working within SAIF to meet the conformance criteria.
Ron Parker: That is the issue?
MOTION: Make the declaration to TSC to allow the documents to go forth as draft for comment, with the provision that the context would be provided before it goes DSTU.
John Koisch: Need to further define through e-mail, to take to TSC for next week.
Ron Parker: Dale/Paul there needs to be dialog with proponents before the TSC meeting.
Paul Knapp: the decision is rational with respect to the organization. It still has the majority of the value, and is a sane and pragmatic solution.
Ron Parker: No, create a structure.
Paul Knapp: We did not want to have surprise in ballot cycle 2 or 3, so we have tried to raise this, and appreciate your time and effort. it is an approach that some people think speaks to the needs they have. Needs look from organization.
Ron Parker: Karen Smith wants a volunteer to help with the SAIF comments.
Adjourned at 5:00pm U.S. Eastern Tony Julian 21:33, 4 November 2010 (UTC)
|Actions (Include Owner, Action Item, and due date)
|Next Meeting/Preliminary Agenda Items|