[New-ITS] Data Types

[Thomas Beale]

Charlie McCay wrote:
>   
>
>> agree they are properties of a number. But when do they arise 
>> in clinical medicine? As far as I can tell, they are there to 
>> model intervals. We say that >5 is 5 -> +infinity. This seems 
>> clumsy, especially since this nullFlavor has significance. I 
>> propose now to toast PINF and NINF altogether and have a 
>> bounded property on an interval
>>     
which is what is done in openEHR as well, just as a point of reference.
>
>   
>
>>> 7.3.1  ED -- OpenEHR mapping
>>> UsagePeriod cannot be dropped as it may contain important 
>>>       
>> information 
>>     
>>> (like how long a reference will be available). I assume that this 
>>> would have to be moved into the archetype during the mapping process
>>>       
>> how important is this? Since the URL cannot be reused, it's 
>> useful but not exactly important. Have you seen it used in 
>> practice? If OpenEHR doesn't model the concept, why would you 
>> try to force it into an archetype?
>>     
>
> It is one of the options for conveying attachments -- to say that the
> image or whatever is available at a URI for a limited time -- I have not
> seen this used in practice, but have heard folk say that it could be
> useful.  If it is dropped, then I suggest that implementation guidance
> should be provided as to how projects where it is populated should
> behave.  Alternatively it should be dropped from HL7 on the basis that
> having been there for a while, no-one has used it, and any conceivable
> use case could be addressed by contractual agreement (which would be
> needed anyway to ensure that appropriate behaviour happened when a URI
> that was needed went out of date).  The recommendation in the wrappers
> document was that the business ack should only be sent when all needed
> URIs had been dereferenced (where they were not persistent).  There is
> still the issue of how the receiver knows that they are dealing with a
> persistent URI or not -- on balance I think that this is something that
> should be in openEHR
>   
I agree that there could be a need to say when a URL might be valid (in 
the future), but I don't think this is part of the URI model itself - it 
is likely to be highly contextual, since you could just as easily have 
no need to say such a thing. On the other hand, saying when when it 
_was_ valid (in the past), e.g. as is done in citations in papers, is 
something I had not thought of modelling, but could in theory always be 
a valid thing to say. If everyone agreed that all models of URI should 
have that, I would probably raise the CR in openEHR.

re: GTS, sending in GTS syntax string format has to be better doesn' t 
it - it means the attribute is just a String and it allows the syntax 
definition to evolve (as long as it is backward compatible) without 
breaking the data, and possibly not harming older software too much.

- thomas beale