April 17, 2008
11 AM EDT (4 PM GMT)
Attending: Rachel Ellaway, JB McGee, co-chairs; Ben Azan , Chara Balasubramaniam Bob Galbraith, Jorn Heid, Frank Hess, Debbie Sher, Valerie Smothers, Arnold Somasunderam, and Segun
The group began with introductions. Jörn and Frank are from Heidelberg, and they are developers of the CAMPUS virtual patient system. They both have an informatics background. Chara is the program manager for eViP. They are working on Labrarynth.
Segun is a new technical developer at St. George's. Segun was practicing MD. Rachel is the cochair of the working group, and she is with Northern Ontario School of Medicine.
JB is also cochair of the working group and he is from University of Pittsburgh. Bob is from the National Board of Medical Examiners. Valerie and Ben are both staff. Ben was premed at Hopkins will be a medical student in the fall.
1. Approve minutes of last meeting
The minutes were approved.
2. Testing of exemplar SCORM package
Debbie commented that she is on path to test the most recent scorm package. Ben added that he tested the package and the ADL conformance test suite and that it worked fine.
3. Supported media types
Rachel commented that this has come up several times. It came up at the eViP meeting in Munich last week. Should we constrain media types expected to be supported by a player? We decided to leave that to the player, but it was causing some implementers problems.
Jörn commented that they want to have all media types be platform independent. A player may have problems reading doc files, for example. Pdf should be used. Ben commented that the next step would be to come up with a list of supported media types.
JB asked what do we support now? Ben replied that pg 47 of the specification mentions basic types - jpg, gif, mov, etc. Valerie confirmed that what we were proposing to create is a minimum list of file types supported by a player.
Rachel replied that it gets fuzzy. From a purist view, that's heresy. Given how many systems and approaches, we may need to compromise. What message do we want to send out?
JB commented that it is also a philosophical consideration - compatibility over flexibility. There are so many media types we may not be aware of. It may be more useful to specify which will be supported.
Then agreed, saying that if authors want their virtual patient to work in many environments, they will limit themselves to supported types.
Jörn explained that in campus they support mp3 and mov and normal graphic types. Quick time will cause problems because it is not supported. If there is a minimum, they will implement the minimum. The minimum will encourage people to use certain media types, such as pdf, flash, etc.
Rachel agreed that we can use the wiki to pull the list together. She asked Jörn to email a recommendation to the group.
Chara added that each eViP partner specified what types of media they would bring. They have that list. Would that be helpful? Rachel agreed it would serve as an important starting point, but we may go down to a smaller subset.
4. Xpath and Xpointer
Valerie summarized the current conversation. The virtual patient spec uses XPath to identify specific pieces of data in different XML documents across the virtual patient. This approach was developed based on advice from the MedBiquitous technical steering committee, and Joel Ferrell, senior technical staff at IBM and chair of the steering committee. Jörn had disagreed with the way XPath was being used, saying that it was an incorrect implementation. Valerie reiterated that the technical steering committee's concern was lack of adoption of Xpointer. She has asked Joel to come back to the group with a recommendation.
Jörn commented that with the proposed use of XPath you have to write your own preprocessor to take out the first element, which indicates the root element and XML file, then switch to a customary XPath parser. If Xpointer is used in concert XPath, this is not necessary. The Xpointer refers to the file.
Ben commented that he has been playing with it. The first part of the XPath tells the player which file to run the X-path in. You don't really need to write anything.
Jörn replied that if using X-pointer, can strip that and just use XPath expression as written. This is a more correct approach.
Rachel clarified that one approach is more correct, the other is more pragmatic. She doesn't have a strong sense of what is right. She advised that we the issue take back to the TSC with better set of issues so that they are better able to advise.
5. Extensions to the base schemas (see examples*)*
Rachel explained that extensions allow the ability to add non-critical elements to MVP data. The concept was reinforced in Miami by Nabil. Campus and CAUSUS need to extend and add specific tags for their systems. One model used by MedBiquitous and other specifications is to add an extension section as a top level element. You could then use ids to cross reference where it should appear.
Jörn commented that they tried using the existing schema without modification. The only possibility was to extend the schema file with their namespace. A written description is on the wiki. With advantages and disadvantages.
The group agreed to add extensions to all 3 schemas and test it out. Valerie and then agreed to work with Jörn on this.
6. Using XHTML for other text elements
The group agreed to keep XHTML use to VPD text and not use it elsewhere.
Rachel commented that she has escaped out characters so that it validates. Jörn commented that if you use images in xhtml, the specification indicates that it should also be referenced by dam. That is a problem if the dam node has image and link to media; it will display two images. Valerie commented that her concern was in tracking learner access to clinically important images that could affect their path through the activity.
Ben commented that we could add something to the specification to indicate that if you want to track an image you must include it in dam. The group agreed.
The group then discussed the use of XHTML in general. Rachel commented that she finds xhtml constraining in OpenLabyrinth. JB commented that his users know what will break things and what won't. It's the compatibility issue. You could have a player that can work in a local system, but when you try to play elsewhere, there would be problems. All of his models require a team. He agreed that to design an authoring system that can handle xhtml is a more sophisticated system.
Jörn commented that they have no problem with XHTML. The authoring system has an html editor. The User can add tables, but no object tags, etc. They can produce xhtml without problem.
7. Recommended subset of XHTML*(includes integrating text and images)*
Jörn commented that <A> is missing - links to outside resources. JB recommended revising the list by email. Rachel commented that the player has to be a web browser if using full xhtml. If you use a subset, other types of players can parse use the formatting.
Jörn commented that there are attributes that might not be supported. That should be specified too. Rachel commented that we will work toward a subset and sign off in Baltimore. Ben will email people to remind them to contribute.
8. Changes to triggering
The group will discuss this in Baltimore.
9. Element name changes (id to path)
The group agreed to change element names from id to path, which is more accurate.
10. Review draft poster
JB will send comments on the poster.
Rachel recommended taking the eViP profile of healthcare lom to the learning objects working group.
- MVP will specify a minimum list of supported media types
- Schemas will be extensible
- The group agreed to keep XHTML use to VPD text and not use it elsewhere.
- Element names will change from id to path
- Debbie will work on testing the sample package
- Jorn will circulate a proposed list of supported media types
- Valerie and Ben will re-examine Xpointer and XPath with the technical steering committee.
- Ben will add extension capability to schemas and work with Jörn to test.
- Ben will change the specification to indicate that if you want to track an image you must include it in dam and not reference it within an XHTML div tag
- The group will comment on the subset of XHTML tags by e-mail
- Ben will change element names from id to path in the schema and the specification
- Valerie will follow up with Rachel and JB regarding the agenda for the VP working group meetings at the annual conference.