MnM Minutes WGM 20100519Q3

From HL7Wiki
Jump to navigation Jump to search

Back to MnM Main Page

Attendees

Grahame Grieve, Woody Beeler, Lloyd Mackenzie, Rene Spronk, Victor Chai, Sarah Gaunt, Gaby Jewell, Ravi Natarajan

Agenda

Object Nets and Object Identity

Discussion concerning Managing Object Identities. Object identity

Outcomes:

  • Without an identifier on an object, it's very difficult to manage the object. So identifiers should never be constrained out
  • even if the static model constrains an association out, it may still be actually in the object model because of context conduction or past exceptions
  • Role.id is a muddle - not clear what the scope of the identifier is. R2 differentiation of business vs object identifier is helpful but does not resolve the problem. Rene will update Object identity to describe R2 on this matter. Recommend pushing adoption of R1.1 as much as possible in this regard
  • Entities - some are identified by code. Propose that Vocab create an property of a code system to say whether a code is also an identifier (Appellation)
  • Participations and ActRelationships - do not have identity. In general there should only be one association between an act and a (act|role) with the same type and effective time

Safe Interpretation of RIM Data

See Safe querying of a RIM-based data model (more RIMBAA focused) and Safe interpretation of subsets of data

  • Lloyd - Short answer: anything that's required in the applicable model.
  • More general answer: in the absence of a conformance context, nothing can be ignored. You can only ignore things because the conformance statement instructed the RIM author that these things must be ignorable
  • It's generally safe to interpret an act in the absence of it's context because of moodCode - except for moodCode Opt (if the mood code is opt, the act is incomplete and the owner must be considered)
  • You always have to consider the intent to decide how the act must be interpreted. Recommend to consider the actRelationships that point at the act - depends on your intent, and consider the type they have as to whether the context they suggest is relevant to your intent
  • It's really hard to interpret any data in the absence of the specification because of the poor quality of our existing models. This is especially apparent in CDA R2 profiles where the choice of Act Relationships is so limited
  • you must consider associations established by context conduction
  • We can't advise on what could be generally ignored. It does depend, but there's nothing that we can say for sure that can be ignored

Back to MnM Main Page