Re: CANCELLED Oct 26 Teleconference Agenda

I'm not 100% certain the action item is related to this, but during the 
F2F I had asked the Interest Group for opinions on the various 
ways of modelling 'controlled values' for a property:

-  as a closed set of literals
-  as an enumerated class (nominals / owl:oneOf)
-  as a class constructed by union (disjunction / owl:unionOf).

It was in the context of the bypass surgery usecase [1] and how you might 
model a controlled list of vessels.

The example was 'Coronary Artery stenosis is a class of heart disease
associated with one of many coronary vessels - each of which with a specific anatomical location':

Class(CoronaryArteryStenosis complete

       intersectionOf(
          ...
          restriction(associatedArtery allValuesFrom CoronaryArtery)
       )
)

I. Nominals (where each vessel is an individual and an instance of 
CoronaryArtery)

Class(CoronaryArtery complete
        oneOf(
          LeftAnteriorDescending
          RightCoronaryArtery
          LeftCircumflex
          ....
        )
)

II.  Disjunction - where each vessel is an instance of the unique kinds of 
Coronary Arteries

Class(CoronaryArtery complete
        unionOf(
          LeftAnteriorDescending
          RightCoronaryArtery
          LeftCircumflex
          ....
        )
)

The former is more amenable for expressing & reusing controlled / coded 
vocabularies (such as HL7 for instance) and easier to implement visually 
as a pick list (for data collection purposes), the latter allows you more 
flexibility to use foundational ontologies such as FMA [2] 
to be precise about the anatomical relationships and descriptors. 
However, what you gain in a more precise description, you lose (may?) in 
uniformity.

[1] http://esw.w3.org/topic/HCLS/F2F/Slides?action=AttachFile&do=get&target=CABGIndicationSlides.zip
[2] http://sig.biostr.washington.edu/projects/fm/

Chimezie Ogbuji
Lead Systems Analyst
Thoracic and Cardiovascular Surgery
Cleveland Clinic Foundation
9500 Euclid Avenue/ W26
Cleveland, Ohio 44195
Office: (216)444-8593
ogbujic@ccf.org

Received on Thursday, 26 October 2006 15:26:59 UTC