CIMI MTF Minutes 20130307
CIMI Modelling Taskforce - Meeting Minutes
Michael van der Zel
- Weekly News & Updates
- CIMI Reference Model Review:
- Summary of Last Week's review of Participation Model
- Datatype changes:
- URI extended to PLAIN_TEXT
- Temporal Types
- Review outstanding tasks
- Lab Results - ADL authoring
- Planning for Leeds meeting & future calls
- Transformation tooling - next week?
CIMI Reference Model Review
Outstanding Tasks for CIMI RM work:
- Clean up the models (e.g. remove unused classes, enumerations etc)
- Create Documentation & Walkthroughs of the model for review and align with Glossary (Jay)
- Write Guidance on use of CIMI RM e.g.:
- Choosing the correct entry point into the archetype
- Documentation on how to interpret Cardinality, optionality, mandatory, assumed values, NullFlavor, default values
- Generate BMM or XSD, depending on the tool we will use
- Finish deliverables (EAP, XMI, BMM, PDF, ...)
- Comparative analysis against Reference Model Requirements
Lab Results - ADL Authoring
- Update from Joey and Patrick
Topics for Leeds Meeting
- Update from CIMI Modelling Taskforce [Linda by phone]
- Reference model finalisation [Michael]
- Laboratory results specialisation
- Terminology Binding work
- Mindmap -> ADL Creation
- Demographics modelling
- Immunization model planning
- Update from CIMI Tooling Workgroup [Harold]
- Proposed architectures
- Terminology Tooling findings and recommendations
- Modelling Tooling findings and recommendations
- Other tooling topics and issues (e.g. CodedText in ADL Workbench, URIs)
- Update from CIMI AML Workgroup [Dave Carlson]
- Review of all Lab Results CIMI Models (ideally in ADL Workbench)
- Review of Lab Results Terminology bindings
- CIMI Enterprise Architecture - Bo Dagnall
- Future modeling priorities
- Future terminology priorities
- Documentation and training (e.g. Style Guide, Reference Model etc)
- Software development tools - decisions on how to proceed
- Software repositories - e.g. ADL and RF2+ repositories
- Software servers - e.g. CTS2 interfaces + Clinical Model server interface
- CIMI Model Web Portal - e.g. for browsing and review of both clinical model and bound terminology content.
- Other Software Apps - e.g.
- ADL-to-AML & AML-to-ADL
- ADL/AML-visualisation apps (e.g. ADL/AML-to-HTML/Mindmaps/Word/Excel),
- CIMI-model-validator (including ADL/AML and terminology validation)
- ADL/AML-to-XSD (via flattened-model),
- ADL/AML-to-GUI (for instance creation),
- Java-based Ecore model of the CIMI RM & Archetypes
- Voting on:
- Final reference model
- Tooling recommendations
Topics for Future Calls
- 14th March:
- Model Transformation demos (Spain, RFC)
- Finalise demographics
- Invite Bo Dagnall?
- 21st March:
- Review Lab Results Specialisation
- Clarify methodology for Terminology bindings & value sets
- Review of CIMI models in ADL
- 28th March:
- Terminology bindings & value sets
- Immunization planning
- 4th April:
- Planning for Leeds
To Do Before Leeds:
- Finalise reference model (including documentation, EA, XMI and BMM) - [Michael & others]
- Finalise demographics modelling [Linda]
- Create Laboratory Results Specialisations [Linda]
- Create main terminology value sets for Reference Model, Lab models and patterns?? [RM: Michael, Lab Models & patterns value sets: IMH?]
- Agree on terminology bindings for rest of Lab Results Models [Daniel/Rahil]
- Add descriptions to all nodes of the Mindmaps [Rahil]
- Document in CIMI Terminology Style Guide [Rahil/Linda]
- Create ADL from Mindmaps for review and discussion [Joey & Patrick]
- Be able to view CIMI models in ADL workbench [Joey to coordinate with Tom]
- Develop a plan for Immunization model development
- Style Guide?
Linda: [goes through today's Agenda]
Michael: I will be available for one hour.
Linda: OK - let's get started.
Michael: Want to repeat something from last time - maybe Mark said it... Want to get Reference Model ready. If Reference Model leaves something open but is transformable, then... finished. Want to make it stable first. Last time - discussion on a number of things... If participation on all entry... cluster... items... Proposal for 13606- only Compositions and Entry.
Linda: I think we moved to allow participation on Sections and Clusters and Entries. We had Use-cases for it. So... would keep since we are using.
Michael: I agree... We re-visit if we...? the number of models. Now I won't change anything.
Michael (cont'd): Some discussion on how to bind terminology and value sets on model. In Reference Model - create reference to value set which will be in external value... XML file...
Linda: Is goal to put in Terminology Server?
Michael: I have (?) and will remove these... So you will know which value set you want to bind.
Michael: The values are not in the Reference Model - are in a terminology server.
Michael: We had a discussion about removing the composition territory, but I don't know if we decided to remove.
Linda: Tom said something about this - the country in which composition was authored.
Michael: One is the language and the other - the country.
Linda: I am still not sure if we leave in Reference Model.
Mark: You at least need a transformation to it since several countries have more than one language and important for clinicians... you can't assume. Can't be one-to-one with country. Certain legal rules in country... to adhere to.
Michael: In model - just a territory ...(?) and not an entry.
Linda: Yes - that is the way in openEHR.
Mark: Where the clinical encounter took place... Some countries also have territories and some don't. Where the legal... is for that patient... and the language... crosses between patient administrative and clinical.
Mark: You can say jurisdiction.
Linda: Michael suggesting it might overlap with... Don't want it to appear twice and contradict.
Mark: A part that is important for patient-care... legal rules. If we need this clinical model... borrowed or used from country. Have to be careful... Use-cases for both.
Linda: Required for safe interpretation of any record.
Michael: Something we rename to jurisdiction.
Mark: Yes - that is my suggestion.
Michael: Have to make sure the distinction between...
Linda: Yes - Other strong opinions on this?
Michael: OK. No other comments, so I'll move on. Data Value Type Changes, a Use-case - wanted to specialize Text to a URI. Was a Use-case by Stan.
Linda: Yes - telecommunication Value... plain text and URI... Stan suggested have a way to allow both.
Michael: I'm sorry - I forgot the context of this thing.
Linda: So would have to say "URI or Text". Stan suggested a URI that refers to Text that could be URI format. Do you have that model there?
Michael: My suggestion would make URI a specialization of plain text. Could have mappings.
Linda: I thought you were going to have... I thought I saw a version of that...
Michael: Yes - string-value and text... URI...(?) I'll check later... would accommodate the Use-case of Stan's.
Linda: So, question is - what happens with temporals below - data-strings... String value stereotypes.
Michael: My reasoning... the temporals are all string-based. Complicates the model now. Makes no sense. There is no Use-case... to do this. OK. I'll move the string-value to...
Linda: And string-value would be an abstract? Could use in models, but not in instances? So would become yellow.
Linda (cont'd): Comments?
Michael: I'll move on... Time flies when you are having fun! The thing that bothered me... temporal discussion... ?402 specifies how string could be used... Why not... [can't hear] string value that has to conform to ISO 8601. To a date... has to be a... So I looked up the different types. In 8601, you have...
Michael: Can you move up... What do you think?
Linda: One thing to add - duration - moved on to quantity... value in units. So... duration... a quantity?
Michael: Maybe because of last week we said as long as things are transformable, stick with what we have now... But ISO 8601... add to Ref Model and... Does it make sense?
Linda: The area that I am uncomfortable with... In clinical model - want to create an element... say it is a date-time. But people may want to constrain down to date. But date is not a specialization... model. So - sibling relationship not clear.
Mark: One approach - to have a data-time and allow it to be constrained at the local level to a date.
Linda: Yes - but Tom's point is adding Date and Time to Ref model? ...not have to specify... So what Michael is saying now... (?) Probably need to remove... value from date and time because inherits ... Galen?
Galen: I was reacting to... Date might be imprecise. Known to only a month or decade. Might not need to constrain further.
Mark: I am in agreement with that. Might be a constraint at local level - part of date-time. So if '00' - say is a month, or use your month. Not know how you say decade without adding (?) to date. Could add a precision to allow that. Other option - if precise to a decade, that becomes a date-time interval... so have to allow for interval.
Michael: That brings us back to - in Reference Model, stop at temporal... ISO 8601. And in clinical model, have to... ISO 8601 allows you to do days and...
Mark: Yes - that is a clearer way of saying this.
Michael: What do you think?
Linda: I still feel uncomfortable with creating models and showing... We had decided to pull duration out. We added duration string... I kind of like it clear-cut in model, but I can see why...
Michael: You can still...
Linda: Not for data-type. With date-time in model, you can still add constraints as you would on the temporal. Some... which will make the model easier to understand. My thought is you could do the same with this. I am curious - why not remove (?) from the model?
Michael: Because ISO 8601...
Linda: OK - will have to add to class. Temporal will need to add a value. Constraints on that.
Galen: Why do we need the temporal since only has...?
Linda: Good question.
Michael: Don't need anymore.
Mark: I think you will find places where people will want to put.... Easier to understand if have duration or... subpart of date-time. So one place in code to go to when deal with date-time...
Linda: So you think will want to...
Mark: I think it is important for the model to be able to have time interval.
Linda: And you don't think the (?) class supports that?
Mark: (?) is a type of time interval... Makes more sense to do in ...(?) model. If we have to explain to clinicians, we have to do a mapping in model. Will not understand anyhow unless had training.
Joey: We had this discussion in Netherlands. Duration is a quantity of time - like 20 seconds. Even though a quantity of units... 3 months and 2 days... similar to a baby weighing 5 pounds and 2 ounces. Interval is different quantities - has start point and end point.
Jay(?): Now date-time looks spurious.
Mark: I think you need a date-time because... Also, as Joey just explained... But needs to have it... for when clinician is describing.
Joey: Previously - for intervals - modeled in the model. A start and end time. When using an interval... in different use-cases... the start-time is not the same start-time in different...
Stan: I just joined the meeting. I'll listen until I catch up.
Mark: I would argue - that makes sense, Joey. Just as we said... time interval... can do as duration...
Linda: Michael - can we come back to this temporal... date-time? Not keen on how it is now. One thing we could do is... move... down and...
Michael: But not have option to express (?) and intervals in ISO 8601.
Linda: But we have ... text.
Stan: I like what you are doing as suggested by Linda.
Michael: We will add... as constrained to ISO 8601 date-string and ... to ... time-string. So I think we have all the Use-cases.
Linda: Another... All enumerations and value-sets in model will be changed to (?) and until we have that, Tom suggests we store in BMM(?) file. So - that is consistent with you suggestion.
Stan: Long-term goal of getting in server and need to get in XML file until terminology server.
Linda: Stan - what (?) either a URI or plain text? I think you suggested... was a type of text or (?) string.
Stan: I don't know if I said, but I agree.
Linda: Perfect. Michael, with value on...
Michael: We need to talk with Tom about encoding in BMM, so discussion with Tom.
Linda: Yes - that is what Tom said.
Michael: I sent you a list, Linda.
Linda: List said... Want to run through each of models. I am in... whether we need supporting diagrams. Can you run through each model?
Michael: Still have to walk through models and see if any dangling classes in these... They can be removed.
Linda: Yes - bunch of stuff deleted from diagram, but not classes - need to be cleaned up.
Michael: Need to... finishing off model...
Linda: I have list of "finishing off tasks". Run through diagrams first and then "finishing off tasks".
Michael: Each diagram...
Linda: No - Null flavor - need to represent instances of our models. Tom raised point that they have archetypes that... Any comments?
Michael: I'm OK.
Linda: Also - change (?) to jurisdiction. The only change. Are all OK? And I suggest we send out over email so all can get one last look and we can vote on as a community. OK? All ready to move on?
Linda: So - except to... If also you could move type about details because details should be last.
Linda: Thank you. I like how simple this mode is.
Linda: OK - data-value.
Michael: So - we changed today - string-value... URI... and simplification of temporal. That's it.
Linda: And also added documentation about constraints.
Michael: Yes - cleaning up the model.
Linda: Any other comments on question on data-types?
Michael: I have one. ... real-time attributes have been removed, because CIMI model is 1-transform away from...?
Linda: For me, you need the data to (?) an instance of model. Integrity check - as long as not a (?) for people... constrain...
Linda: Does anyone know when might want to constrain the (?) integrity check on multi-media?
Michael: So - I'll remove.
Linda: Other questions on data-type?
Michael: Supporting classes.
Linda: Are you using this? Not using party-name - using direct reference? I can't think of... where we are using these. It would be ideal if we could remove these. Consolidate things nicely.
Michael: ... not being used.
Linda: Perfect - one less diagram for people to understand. Are all OK?
Michael: I agree.
[diagram: UID at the top, and UUD, ISO OID...]
Linda: Can we move into...?
Michael: These are primitive types, so...
Linda: So not have to be part of model?
Linda: Perfect - down to 3 diagrams.
Michael: So we'll clean up. Down to...
Linda: Any final comments on Reference Model? We'll clean-up and send out. Other comments today? Let us know.
Linda (cont'd): So - this is the list you sent me, Michael. #1 - to clean up models (remove unused classes, enumerations, etc.)
#2 - Create Documentation and Walkthroughs of the model for review
Linda: What is a Walkthrough?
Michael: I found useful from HL7... Walkthrough. The whole model explained in...
Linda: OK. Jay - want to talk to documentation-Glossary?
Jay: Not a lot to say, but we'll be working on it in the next week or so.
Linda: Next is Write Guidance on use of CIMI RM. What... Michael?
Michael: Probably in Style Guide. Cardinality, optionality, mandatory... ADL has some way of expressing what is default value... What choices are in our models...
Linda: OK. Now - Null flavors. Part of same list - documentation on how to interpret?
Linda: Next - Generate BMM or XSD depending on the tool we will use. And finish deliverables - EAP, XMI, BMM, PDF... Everything, Michael, you want to cover?
Michael: Yes. Jay on Glossary. Writing documentation is hard for me, but I will do my best.
Linda: Calling for volunteers.
Linda: Yes - volunteer!
Jay: Actually it is Sarah. [Jay is using Sarah's sign-in for gotomeeting]
Michael: We have... needs work.
Linda: Not do till model is cleaned up. Thanks for your help with that, Jay. And thanks, Michael. Great work. Anyone have other things to add on this topic?
Linda: In next few weeks, Dave, you'll see XMI coming your way... Can help...
Dave: Yes... bugs in Sparks... No (?) in models in Eclipse. Because improperly represented in EA.
Michael: Yes - I have as a note - Harold...
Dave: Yes - We'll talk tomorrow - Michael, Harold and I. Possible to write a (?) script to put the documentation in the model.
Linda: Thanks, Dave. Thanks, Michael. That wraps that up.
Linda (cont'd): Next - Joey and Patrick - ADL.
Joey: We just got started. Now - building up a parser... Mindmap files to generate ADL. I hadn't seen the bullet point. I thought Mindmap to ADL converter. But now... We would generate existing map... to ADL. But are you thinking... the LOINC - those generated as well?
Linda: Have not had a chance to go through. I thought you were manually authoring ADL.
Joey: Yes - more fun... [laughter]
Stan: It's all about you, Joey! [laughter]
Joey: So - you wanted... with difference?
Linda: We have four more meetings before Leeds meeting. Will we have something by then?
Joey: Yes - I think we can have by then.
Linda: Would be great to show Lab Results in workbench...
Joey: I know... new release... All builds are done... and installer today. So by this weekend, he was going to give us a version.
Linda: Great - and if you could start testing to see if... And also with Michael, who is... BMM?
Dave: Also, Joey, if you have parser for Mindmap... lean on you to... ADL...?
Linda: Also - if you have difficulty in parsing Mindmap because something I am doing, tell me so I can stick to formatting [however] you need.
Linda: Thanks, Joey and Patrick. So - next is Leeds meeting. Virginia asked us to brainstorm for MTF to do some work over there. We have 4 meetings - 14th, 21st, 28th of March and 4th of April. So, want to discuss where we want to be before Leeds, and how to get things done in meetings before then. Must be an update on CIMI MTF. I probably won't be able to attend Leeds meeting. Michael - useful for you to present the Reference Model. I would like to vote at meeting... OK, Stan?
Michael: [can't hear] ... week before?
Linda: Yes. Are you happy to present there, Michael?
[Linda show slide - Topics for Leeds Meeting]
Linda: So - other topics. Laboratory Results specialization. Last time - patterns and... this time, specialization... Start to spend more time on this once tooling is... The Mindmap and... Love to demonstrate the ADL being produced. The demographics models. Immunization models.
[Refer to slide - Topics for Leeds Meeting]
Linda: Update from CIMI Tooling Workgroup; Proposed architectures; Terminology Tooling findings and recommendations; Modeling Tooling findings and recommendations; Other tooling topics and issues; an Update from Dave on CIMI AML Workgroup; Review of all Lab Results CIMI Models; Review of Lab Results Terminology bindings.
Linda (cont'd): Planning; Future modeling priorities; Future terminology priorities; Documentation and training; Software development tools
Linda (cont'd): Software repositories - not sure where things are with this - Stan? Are you ready for the Leeds meeting?
Stan: Yes - put on the to-do list.
Linda: OK. Software servers; CIMI Model web portal; Other software Apps... ADL - visualization; CIMI-Model-validator; ADL-to-XSD - requires flattening; ADL-to-GUI, ADL-to-XML...
Linda (cont'd): Great to do a vote on final Ref Model... A bunch of ideas.
Michael: About software Apps - assume only from ADL to something else. Also AML.
Linda: Yes. First should be ADL-to-AML and AML-to-ADL. All of these should be both. So, ADL/AML-visualization apps, etc.
Dave: To add. Generation of Java-based Ecore model. Hopeful we'll have a proof of concept. ADL... AML... to Java-based Ecore.
Linda: CIMI Ref Model?
Dave: Plus all of the archetypes that would allow you to read and write instances.
Linda: Great! Once we get (?) in repository, need to look at software Apps... in distributed system so we can all work in. Anything to add to the list? Dave - nothing else to add? Can you attend the meeting?
Dave: No - no budget to attend, but I'll dial in.
Joey: Bo had mentioned... "I don't how much I can contribute, but can help some".
Linda: Yes - need to think about meeting before the meeting and at the meeting, because Bo had mentioned... I think a good idea, but others need to agree. Alternatively - invite Bo along to meeting, or have a smaller group. Stan - are you happy to proceed with Bo?
Stan: Yes - I can contact Bo. Last I saw, someone was going to...
Linda: Yes - an email since. Bo said he could meet the weekend after HIMSS. We can meet, or invite him along to MTF meeting.
Stan: Yes. I am not an expert in this area, but my intuition says this will help.
Linda: Thanks for reminding me, Joey.
Michael: Bo said he was willing to meet.
Linda: We could discuss offline. Let him know our thoughts. Yes. We'll follow offline.
Stan: On Thursday the 28th, I responded to Bo and said to set up a time. I probably waited to respond to all, but mistakenly responded only to him.
Linda: So - if you can let us know.
Stan: Yes - I will resend to all right now.
Linda: Great. So we can prepare beforehand or just invite him to the meeting.
Linda (cont'd): Thanks, Joey. Mark - any feedback on agenda?
Mark: I probably won't be attending the meeting except by phone, if available.
Stan: Yes - certainly we will have the ability of attendance remotely.
Linda: Michael - anything you want to add to Leed's topics?
Linda: If you do, please let us know. Mike Lincoln - no comments from you?
Linda: Patrick - anything?
Joey: No - this is Joey.
Linda: Jay - anything?
Linda: And Stan - anything there that shouldn't be?
Stan: Looks great to me.
Linda: Great - if you have any ideas after the meeting, let me know. So, as I said, we have 4 more meetings before Leeds. Have to work out how to best use the next 4 weeks. If I have enough people to work on... I'll go over the list. So - finalize the Reference Model. Michael - I have you down with help from others.
Michael: Another... [Reference Model] Requirements...
Linda: Good point. I'll put "Comparative analysis against Reference Model Requirements" [under CIMI Model Review].
Linda (cont'd): Next - Finalize demographics modeling... Party, actor... If we could get some ADL generated, might make the review easier. Michael - do you need meeting time for Reference Work?
[no response - Michael has left the meeting]
Linda: I'll assume most is off-line work. Next - create Lab Result Specializations - Linda.
Linda (cont'd): Next - Terminology. One is "Create main terminology value sets for Ref Model, Lab Models and Patterns" Rahil has offered to [help with Terminology], so I'll put her down. Anyone else with Terminology background and can help?
Linda: Michael left - I will put him down for Ref Model.
Stan: I will see if anyone at IMH who can work on it.
Linda: Thanks - need to go over and see what value sets we need.
Linda (cont'd): Next - Agree on Terminology binding for rest of Lab Results Models - I'll talk to Rahil. Rahil has offered to add descriptions to all nodes of Mindmap.
Linda: Next - Create ADL from Mindmap. Joey and Patrick. Next - Be able to view CIMI models in ADL workbench - Joey coordinate with Tom?
Joey: Well - I said...
Joey: I'll send him an email.
Linda: Great. If you need me to follow-up... Next - develop a plan for immunization model development. Would be good to go to Leeds with a plan for immunization. Stan - what priority do you think?
Stan: First step - ask... primarily the DOD and VA were working on immunizations. First find out if any timeframe... If we do sooner so they could use our work... we'd make it a priority.
Linda: Who do we talk to?
Stan: Mike Lincoln and Steve Hufnagel.
Linda: I'll send an email to follow-up on what time-frames... OK - Style Guide... Don't know if we should aim to have any of this before Leeds. Might be too ambitious...
Linda (cont'd): Next - how to use our time. The 14th of March - Model transformation demos (Spain, RFC). Anyone still interested?
Linda: Good. Also - Results4Care presentation of transformation-related work. Stephen has said NEHTA, but not ready yet. So if any more, let us know.
Linda: Also - finalize demographics. Also, review Lab Results specialization, and review what Joey and Patrick are doing in review of CIMI Models in ADL - on the 21st of March. And on the 28th of March, Immunization planning. On the 4th of April - Planning for Leeds.
Linda: I'd love to add more topics. Also, depends on if have terminology meeting on Tuesday. Any other suggestions for future meetings?
Stan: So - any binding discussions would be on Tuesday?
Linda: Yes - usually on Tuesday, but since doing Tooling on Tuesday, maybe Terminology discussions on Thursday. Would also help lab result...
Stan: Yes - we've said things... I understand conceptually... either work that Rahil is doing or we do together to understand how all that tangible is ADL and AML...
Linda: Yes - I spoke with Tom about Scottsdale... need decisions on whether we are progressing in that way.
Dave: I'd like to second that request. Clarify terminology bindings... Looking at Mindmaps... how object relationships used on... When I talked to Daniel, incorporate semantic triple into ADL. Important to get... how object assigned to relationship and relationship to object.
Linda: OK - important to have on Thursday meeting. OK - that is all I have for today. Anything else? OK - we can finish early. Thanks for attending.
[end of meeting]