FYI, there are two upcoming W3C workshops that may be relevant to this discussion, on how to use RDF/OWL for application integration:
http://www.w3.org/2011/09/LinkedData/ http://www.w3.org/2011/10/integration-workshop/ (The first one would focus more on a semantic web approach than the second one.) - Dave From: Hau, Dave (NIH/NCI) [E] Sent: Sunday, September 18, 2011 6:47 PM To: Kerstin Forsberg Cc: Jim McCusker; conor dowling; [email protected] Subject: RE: FW: A Fresh Look Proposal (HL7) Thanks Kerstin. The DCM spreadsheet is informative for me. There are several clinical model sources I had not known previously. I think the DCMs or archetypes could potentially be defined as graph patterns via SPARQL queries, involving concepts from standard domain ontologies. These DCMs could be resources exposed by REST as documents in sections according to the hData record proposal. In addition to that, the SPARQL endpoint could be a query resource in the REST service as well, supporting more granular queries or for other graph queries, under an access control mechanism. The SPARQL query for a DCM would provide a level of semantic validation for a submitted RDF. Further validation could be via the Pellet Integrity Constraint Validator (ICV) for example, involving reasoning on the data definition together with the value set as we discussed before. Any other thoughts or comments are welcome! - Dave From: Kerstin Forsberg [mailto:[email protected]] Sent: Sunday, September 18, 2011 6:21 AM To: Hau, Dave (NIH/NCI) [E] Cc: Jim McCusker; conor dowling; [email protected] Subject: Re: FW: A Fresh Look Proposal (HL7) 2011/9/18 Hau, Dave (NIH/NCI) [E] <[email protected]<mailto:[email protected]>> Jim, I was thinking more about units of content as in a container storing properties from different classes, for a clinical purpose. The classic example here is when an EHR needs to store or report a blood pressure measurement, what are the various data elements that need to be reported to provide enough context to interpret the reading. Here's a nice overview of Detailed Clinical Models https://docs.google.com/document/d/11xGZs4drfwb7TY15Iwpk52UIxZcFhqMcMp1i3yV4To8/edit?hl=en_US&pli=1 For the classical BP, and other Vitals Signs examples, it would be interrsting to see see a real example of clinical data using the BFO based Vital Signs Ontology http://bio-ontologies.knowledgeblog.org/155 Kerstin Forsberg AstraZeneca
