Friday, October 29, 2010

CCD and VMR

I missed the HL7 VMR call last week because I was somewhere else.  The discussion that week was on how to procede to fill in the class models.  One suggestion proposed was to use the work that had already been done CCD, but apparently others felt that they needed to go back through HL7 Version 3 models and constrain them down to what should go into the classes, following the HL7 process.

So, I was surprised to see that proposal, because from my perspective, we have already been there and done that.  The next step should be to adopt that work that was done in CCD using the HL7 process and methodology and start working on subsequent steps.

The real challenge from my perspective, is NOT the static models.  We can build static models until we are blue in the face, and the HL7 tooling will give us, depending upon your point of view, beautiful or deadly ugly XML schemas.  But static models don't tell the real story behind what the VMR needs to do, and it's not about the XML.  In fact, its so not about the XML that the original draft put that out of scope.  The reason being that the VMR should be a way to access information that comes from any number of sources, including HL7 Version 2 messages, Version 3 messages, CDA documents, and even NCPDP SCRIPT and X12 messages.  It's not about messaging, which seems to be a difficult concept to get away from because the HL7 development process seems to bring so much attention to the messages and models associated with them.

Given a particular piece of knowledge, the VMR needs to be able to navigate to clinical statements accessible from there, and to be able to filter and locate other knowledge that needs to be attended to.  This is a model of dynamic behavior.  HL7 is still working out some of the dynamic behaviors, and I'll note that dynamic behavior is NOT about message content.  It's about what happens between systems and the responsibilities between them.

In the longer term, it points to another problem that we need to address, not in VMR, but through HL7, which is how to advance our methodology to bring attention to dynamic behaviors.  I'll just put it on my punch list, and maybe someday, I'll be able to punch out.

0 comments:

Post a Comment