Re: ISSUE-5: n-ary datatypes - use cases ISSUE-53

On Nov 12, 2007, at 2:14 PM, Jeremy Carroll wrote:

>
>
>
> ISSUE-53
> http://www.w3.org/2007/OWL/tracker/issues/53
>
> is also a relevant usecase for n-ary datatypes.
> [[
> I need the ability to say something like:
> Patient that (testdate > enrollmentdate + 60)
>
> Was trying to represent the following inclusion criteria in a  
> clinical trial:
>
> The patient must have negative gadolinium based MRI of the  
> contralateral breast, no more than 6 months prior to study entry.
> ]]
>
> I am curious as to whether this can be met without defining a class  
> of patients with this property. e.g. is preprocessing adequate, or  
> maybe a posthoc SPARQL query of some results - rather than a  
> declarative definition of a class of patients based on a numeric  
> computation.

The test date and enrollment date will typically be recorded as  
property values of instances. Test date will be something like an  
instance of test associated with a patient record. Enrollment date  
would be some value off an instance of study. So this *isn't* the  
specified cases where the two properties are of the same individual,  
rather the extended case.

The use case we decided on in HCLS is that where a researcher who is  
about to do an investigation is evaluating patient records to see who  
might be a candidate for a study. In that case the enrollment date  
would be known and the case reduces to testdate < constant.

Based on this, I propose we close ISSUE-53. Should another example  
show up we can open a new issue.

-Alan

>
> Jeremy
>

Received on Sunday, 18 November 2007 06:56:37 UTC