2010-01-12 TSC WGM SAIF Architecture Program Minutes
|HL7 TSC SAIF Architecture Program Meeting Minutes
Location: 2011Jan WGM, Sydney AUS
Time: Q4 - 3:30 PM AUS local time
|Facilitator: Austin Kreisler||Note taker(s): Tony Julian and Lynn Laakso|
|x||Woody Beeler||HL7 FTSD Co-Chair|
|x||Gerald Beuchelt||Mitre Corp|
|x||Keith Boone||HL7 Structured Documents Co-Chair|
|x||Jane Curry||HL7 Tooling Co-Chair|
|x||Grahame Grieve||HL7 ArB|
|x||Rob Hausam||HL7 Vocab Co-Chair|
|x||Wendy Huang||HL7 IC Co-Chair|
|x||Tony Julian||HL7 FTSD Co-Chair|
|x||Paul Knapp||HL7 ITS Co-Chair|
|x||Marc Koehn||GPi, EA IP PM|
|x||Austin Kreisler||HL7 TSC Chair, DESD Co-Chair|
|x||Lynn Laakso||HL7 staff support|
|x||Tony Lam||MOHH SG|
|x||Patrick Loyd||HL7 T3SD Co-Chair|
|x||Ken McCaslin||HL7 TSS SD Co-Chair|
|x||Lloyd Mckenzie||HL7 MnM Co-Chair|
|x||Charlie Mead||HL7 ArB Chair|
|x||Galen Mulrooney||HL7 SOA Co-Chair|
|x||Frank Oemig||HL7 Co-Chair|
|x||Melva Peters||HL7 IC Co-Chair|
|x||John Quinn||HL7 CTO|
|x||Mark Shafarman||HL7 Templates Co-Chair|
|x||Pat Van Dyke||HL7 Co-Chair|
|see sign in sheet|
|Quorum Requirements (Chair +5 with 2 SD Reps) Met: n/a|
SAIF Architecture Program
Meeting was called to order at 15:30 GMT + 10 by Austin Kriesler.
Previously was titled Enterprise Architecture Project Meeting.
Austin presented the goals:
- Identify what projects are necessary to stand up an HL7 Architecure \based on SAIF and demonstrate use of that architecture.
- Project will be demoing - Composite order & CDAR3 to cover all paradigms and deal with cross-model consistency.
Woody Beeler:Saturday you refered to a SAIF framework and HL7 implementation guide.
- Output of program will be an implementation guide.
Patrick developed a spreadsheet that identifies some of the things we need.
Woody Beeler: Neet a higher level of abstraction that is acheivable. We have a definition of artifacts needed. There are other things we need. These don't populate the SAIF, but where does the RIM fit in the SAIF?
Galen: Governance - we govern the RIM, but dont govern the DAM. It is governable - who owns, who documents, etc.
Grahame Grieve: THis is the stack of things that would make SAIF if we had a whole stack that is SAIF.
Lloyd McKenzie:What are the near term things?
Austin Kriesler: We have a blank slate.
Woody Beeler: Not that blank. We know there is a reference model, and governance. DOnt know if RIM is part of SAIF. We ned to list artifacts on which we can focus.
Patrick Loyd: We asked EHR to play, because we need functional profiles. SOme are boxes we have not played in before, and dont apply to other groups. Involving services, EHR, and STRUCDOC are about filling in the boxes.
Woody Beeler: Here in yellow are items of the SAIF, and here in grey are the maybes.
Paul Knapp: Templates - is that a message or a CDA?
Mark Shafarman: Either one.
Patrick Loyd: Hope you can document lab order or cda document.
Keith Boone: Template is a set of constraints that can be applied to a model.
Grahame Grieve: Maybe we need a DAM for the RIM. I am not sure what we are trying to do.
Jane Curry: I am trying to understand.
Lloyd McKenzie: We have a bunch of stuff with the same methodology, maybe we would define the RIM differently.
Austin Kriesler: Scope is to define projects using the SAIF for OO's composite order.
Grahame Grieve:The RIM should be SAIF as well.
Woody Beeler: If you have not figured out how to represent the vocab and others in SAIF you can't start.
Charlie Mead: That chapter is now in draft form.
Lloyd; The SAIF book is not done - not even close, I have only seen two chapters. That does not mean we cannot move forward in a parallel track. We will need to create stuff, most not different from what we have, but needs to be documented. SOme is stuff we wanted to do, but did not.
Woody Beeler: You did not hear serial from me. You have to get those things in as rapidly as possible.
Lloyd McKenzie: Define artifacts, methodology, and Infrastructure reference structures, Content others can use.
Charlie Mead: THat is HL7 SAIF IG, not SAIF book.
Lloyd McKenzie: The instances, and the changes, will happen in stage 3
Austin Kriesler: We have
- SAIF book ballot project
- SAIF implementation guide For HL7
- SAIF Behabvioral framework
- OO Composite order project.
Woody Beeler: Someone said project services is doing that. It does not belong there.
Austin Kriesler: PS has a piece, not the whole thing.
Lloyd McKenzie: Anyone can play in the later stages.
Austin Kriesler: SAIF IG for HL7 needs to be first
Lloyd McKenzie: SAIF IG is MnM, strucdoc, SOA.
Marc Koehn: There is an idealized process. . .
Austin Kriesler: We need to get started.
Marc Koehn: It most go in parallel.
Austin Kriesler: SAIF IG needs to get fleshed out, so the OO and CDA projects can take, and give feedback.
Lloyd McKenzie: My goal would be a list of artifacts by the next meeting.
Paul Knapp: Message? CDA?
Patrick Loyd: Message, CDA, Service.
Lloyd McKenzie: Logical approach depends on teh artifacts, and the methodology. Then figure out the infrastructure artifacts.
- Artifact definitions
- Methodology governnance
- These parts of infrastructure are defined, these are not.
Charlie Mead: The first two are the HL7 IG.
Lloyd McKenzie: They can be constructed sequentially.
Charlie Mead: We need a couple of projects - it is iterative/incremental.
Lloyd McKenzie: Define what we need to build.
Marc Koehn: What is the journey? IF we had that, we could document, track, and manage. The struggle is that the streams don't have to be 100%. If the project is to create version zero, and determine when it needs to be perfect. Lets define the artifacts, and checkpoints.
Patrick Lloyd McKenzie: We spent 1:10 churning on the first step.
Austin Kriesler: Who needs to be at the table?
Woody Beeler: We need EHR and others to think about those. I want to know what the rows are, and what they are called.
Lloyd McKenzie: We need a catalog.
Woody Beeler: What are the rows?
Patrick Loyd: ECCF
Woody Beeler: I dont know what it looks like.
Keith Boone: It is the 3x4 grid.
Lloyd McKenzie: We need to make certain that the introductory power-point is up to date, and make it available to the players.
Woody Beeler: Intro to SAIF ppt needs to have terminology updated for what the names conceptual logical, implementable for 3x4 grid. It is rash to think we should have stewards for each cell.
Lloyd McKenzie: Cells are categorization. Important are the artifacts. Content is on the artifact basis.
Charlie Mead: Placement in the ECCF is not important. But they must be there. Content is most important, second is representation, and third is location in the grid.
Keith Boone: It does not matter where you put it.
Charlie Mead: Least important.
Keith Boone: With different scopes, artifacts may show up in different cells.
Charlie Mead: HL7 needs to define where they go.
Marc Koehn: Within an organization the location of the artifact must be agreed upon and placed consistently. The place where things make sense may derive from how we want to make decisions about it. It would be nice if we had a conceptual notion that would not move. We need to understand how element contents move through the decision practices - and until we do it it won't make sense.
Keith Boone: I have gone through the box-putting excercise. Filling out the box is a dart board.
Lloyd McKenzie: It is not going to be the driver for anything we do - it is a collector.
Jane Curry: When we put it togethere there were reference documents that do NOT go in the ECCF - EHR-FM and the RIM are examples. They are not documents used for conformance. An instance of the specification stack as used for conformance would become part of the ECCF grid. Artifacts in the spec stack develop in the two perspectives of representational grammar and the implementation grammar.
Woody Beeler: Is the RIM an artifact of the SAIF?
Jane Curry: No.
Charlie Mead: When you instantiate an instance of a stack in the matrix, it has a scope. DOes not have to be scoped to a implementation. It has informational and behavioral model. In the HL7 IG, the RIM, core principles are discussed. Does that show up in a stack instance? No, things you develop using the grammar. You take the product development life-cycle, and control the risks by governance. Most will show up in the ECCF stack, but may not.
Woody: notes that governance cannot be applied solely to processes in the ECCF, if RIM and Datatypes are outside the RIM.
Charlie Mead: across the product development lifecycle as discussed by the TSC on Sunday you identify metrics relating to things needing to be governed but they may not be solely represented on the ECCF.
Keith Boone: Lloyd said the cell did not matter. We found that the placement informs the intersections. Characteristics about columns and rows create notation on the intersection of x and y that illuminate
Lloyd McKenzie: It does not drive the process. It merely helps us with the understanding of the artifact. We will figure what artifacts we need, some in single cells, and some in multiple. A person responsible for an individual cell would not work.
Paul Knapp: Can an artifact live in multiple cells?
Charlie Mead: Yes, they are not normalized. As long as you know they are there in an organization’s implementation guide, you are ok.
Keith Boone: CDA R2 lives in several places, within an artifact that section x lived in one cell and section y perhaps in another cell
Austin: we’ll need a program scope statement, and we’ll need a project team.
Grahame: I can see it on a de novo project but to retrofit. He suggests to do some domain analysis models for the RIM.
Lloyd notes that if we started from scratch without the existing methodology the RIM would probably look different that it does now.
Grahame:If the project depends on the RIM, then the RIM needs to be SAIF’ed first.
Charlie Mead: notes that the RIM is the information framework reference grammar.
Lloyd: The SAIF book is not close to done – review on existing chapters has not seen revision come back on feedback. A parallel track can exist with this methodology where we create “stuff” but we can use what we have now if we document it better. Some will be new, some we have talked about but not having gotten around to it formally. Define the artifacts, define the methodology, then define the infrastructure structures in place. Then stage 4 is content that someone other than HL7 might be able to use.
First we have the SAIF book ballot project. Then we have the SAIF IG project. SAIF and Sound is the PSC project Galen asks what about that. Patrick says with this structure of the program management they want to draw in all SAIF related projects to coordinate activities.
Woody: It doesn't belong with Project Services – that they have responsibility for a piece of it but not the whole scope.
Composite order is a stage 4 piece; they are already in progress though they know that it is unstable until the other elements catch up.
The priority now to get the SAIF IG, MnM, Structured Docs, EHR, SOA involved with defining artifacts and methodology.
Marc notes that the parallel effort with Composite Order will inform the activities of the other groups, that it is important to run in parallel.
Austin refers to it as an incubator project for the SAIF IG and can test early if something will work.
Lloyd McKenzie: By next WGM I hope to have a list of artifacts, and where they exist. We may not meet the target, but we may.
Charlie Mead: If you are volunteering ArB to run it, let's say the SAIF book is evolving to have a cut by next meeting. Will it be all the artifacts we need or just categorization of what we have? Lloyd is working on a list of old and new. Charlie asks if he’ll be ok if the artifact grammar changes a bit between now and then for the SAIF Book.
Paul Knapp: Should we map our existing artifacts into the ECCF?
Lloyd McKenzie: yes at a minimum.
Paul Knapp: will there be review?
Lloyd McKenzie: There will be determination of the location.
Charlie Mead: Content is more important than representation.
Lloyd McKenzie: There are questions for an artifact, but there is a template on the wiki being reviewed by ArB and others to be ready tomorrow night. Lloyd and Charlie both say that the location of the artifacts isn’t as important as getting the list (content). Lloyd described a draft template of the 15 criteria that are needed to describe an artifact and is being developed on the wiki and will be available tomorrow night.
Woody Beeler: There was an implicit statement that this is an ArB project. It should be independent of any work group.
Lloyd; The drivers are MnM, strucdoc, Arb, EHR, ITS, IC, SOA, Structured Documents.
Woody Beeler: It should be within a work group.
Patrick Loyd: When someone takes the lead it gets done. I suggested ArB because I had no better answer.
Woody Beeler: I am not from MnM joining an ArB project, we have a project.
Keith Boone: Broadening the participation would not make it happen faster. We need to create, and present back to the wider audience. Create the dartboard and then put it out there for others to throw spitballs.
Jane Curry: I suggest that if ArB is monitoring if not driving, it would be usefull to have a wikipage off arb page with links to other material.
Lloyd McKenzie: There is already a wiki page to manage the artifacts.
Patrick Loyd: That is the next step. I would write up Project scope statement(s), then socialize with the 5 groups and bring it back. We have enough info to frame it and bring it back. It should only take a few days.
Austin Kriesler: I would like to see a listserv for the 'program'.
Ken McCaslin: I'll take the action item to get that set up.
Meeting adjourned at 4:30pm Sydney (GMT +10)
|Actions (Include Owner, Action Item, and due date)|
|Next Meeting/Preliminary Agenda Items|