This wiki has undergone a migration to Confluence found Here
RIMBAA 200909 WGM Minutes
Revision as of 12:32, 23 September 2009 by Rene spronk (talk | contribs)
Minutes of Tuesday Q4
- Attendees:
- Peter Hendler, KP (co-chair)
- Andy Stechishin, GPi
- Rene Spronk, Ringholm (co-chair, scribe)
- Nick Radov, Axolotl
- Davide Magni, ITAL TBS
- Andrea ceiner, ITAL TBS
- Amit Propat, Epic
- Steve Fine, Cerner
- John Timm, IBM Research
- Dave Carlson, VHA
- Robert Worden, Charteris
- Mark Tucker, Regenstrief
- Ravi Natarajan, NHS
- Rene calls the meeting to order at 15:33
- Rene/Peter briefly present the RIMBAA technology matrix.
- Tool Presentation by Dave Carlson and John Timm.
- Primary aim of the tool was to do (CDA-) model refinement in UML, serialize in schema, or back-out and generate MIF (required for publishing).
- Templates can't yet be transformed back to MIF. Rik Smithies notes that templates can be expressed as RIM models and hence as MIF. Lost of template-discussions are still ongoing, thyere is no single way of expressing templates at this point in time.
- Model driven framework for generating run time API, with clone names, with one or more templates validated, consume XML instances that deserialize into appropriate template.
- UML-OCL template model (by standards creator, expressed as a UML Profile, annotations in EMF to express OCL constraints), transform to code generation model (EMF), code generation to create Runtime API in Java (inclusive of the OCL constraints), produce/consume/validate CDA instances.
- Template model: is-a and has-a relationships between templates, given that one template (LIM, e.g. specialization of an IHE template) is a specialization of another template (LIM, e.g. IHE template) is a specialization of a RIM-derived model (CIM). Requires manual work to determine specialization hierarchy (based on textual requirements in CDA implementation guides)
- John: generalizing this to all CIMs (not just the CDA model) is definitely possible. Requires additional work on the MIF to Ecore transformation code.
- Link to the software: mdht.projects.openhealthtools.org, the presentation is available at https://mdht.projects.openhealthtools.org/files/documents/58/217/CDA_Tools_Overview_HL7.ppt
- Primary aim of the tool was to do (CDA-) model refinement in UML, serialize in schema, or back-out and generate MIF (required for publishing).
- Tool Presentation by Robert Worden presents a mapping tool; in preparation we asked him to focus on mapping of a proprietary database model to an RMIM.
- Robert present why we need mapping at all. Mapping is easier than migrating. Publishing shareable [i.e. not hardcoded] mappings (e.g. v2-v3) is a good idea and drives adoption of v3.
- Most mapping issues are related to the mapping of non-leaf concepts in XML. Map structure A to a semantic model, map B to the same model. Use an R-MIM format (MIF transformed to EMF) as the underlying semantic model.
- The tool ultimately generates transformation classes (or XSLT).
- Andrea: ItalTBS adds validation to these classes.
- Robert demoes a mapping from a proprietary database structure to a CMET. Robert has no EMF for 'the RIM', only for R-MIMs. Would lead to a different type of mapping.
- Tool can be found at: HL7 Homebase; Presentation Slides
- Announcements
- out-of-cycle in Amsterdam on Oct.27 2009,
- request for speakers in Phoenix,
- announcement of joint meeting with ITS WG in Phoenix
- Motion to adjourn (Peter/Andy) at 16:59