Re: Linked Data for ICD? OPCS-4?

Jeni,


> I was just looking at the statistics we have on UK hospital episodes 
> [1] which use ICD-10 to indicate the primary diagnosis related to  
> the episode. I'm intending to model the statistics using SKOVO[2].

Just looked at SKOVO. To me, the model really looks like it has great  
semantic "leverage". Also, I see it uses SKOS. I'd love to see how  
you'd use it in practice.

> Unfortunately from what I can tell the NCI Metathesaurus doesn't  
> seem to define URIs that are possible to construct based on the  
> ICD-10 codes. Or it does and I haven't been able to find them. I'd  
> be grateful for any advice.

To the best of my knowledge, WHO has not defined an "official" uri- 
based naming scheme for ICD codes.

There is a registered OID for ICD-10 (2.16.840.1.113883.6.3, see inter  
alia http://www.hl7.org.uk/version3group/oids.asp#Some%20HL7%20allocated%20OIDs%20relevant%20to%20the%20UK 
  ).

Since OIDs are (expressible as) urn's, in effect there is therefore a  
(admittedly, somewhat unsatisfactory, but "UK-official") way of  
referencing the ICD-10 system tout court in a urn syntax.  I do not  
know (and doubt) if anyone has defined a mechanism for attaching  
specific codes to that OID prefix in a urn-based way.

(Note: These OIDs are mostly in use within the HL7 community, and that  
community doesn't really use them qua urn's -- it just uses them as  
unique identifier strings within XML messages defined in their xml  
schema for medical documents). ( i.e. they might e.g. have a fragment  
in a HL7v3 clinical document that looks like this: <code code="J96.0"  
codeSystem="2.16.840.1.113883.6.3" codeSystemName="ICD-10"  
displayName="Acute respiratory failure"/> )

Good news:  ICD-10 is one of the coding systems incorporated in the  
Unified Medical Language System (UMLS) of the US Natl Library of  
Medicine. And I was told by Olivier Bodenreider several months back  
that NLM might/would/hoped to release UMLS in an RDF format sometime  
in the (?near?) future. If so, one could use the UMLS identifier in  
its uri form as a way of referring to ICD-10 concepts. This would be a  
good thing, I think.

I'm ccing Olivier, who maybe can say if that project is still "on".


>> Would you be interested in talking to the HCLS Terminology group (I  
>> chair it) about it on an upcoming Thursday morning call? I predict  
>> you'd get a lot of interest and support -- even help, if you like.
>
>
> If the above still sounds interesting, and provided the timing  
> works, sure.

It's Thursdays at 1400 UTC (1500 UK time). I'll start sending you  
announcements, and if you can be there sometime in the next few weeks  
I'll put you on the agenda.

John