Karsten Hilbert wrote:

On the point about no need for a standard in asking for the
atomic data, as Karsten points out, there are many ways to
ask for "Patient First Name":

PFN
PatientFirstName
PatientGivenName
etc,etc


True.

this is actually where archetypes come in - they normalised names of nodes in data such as this, taken from a reference vocabulary; they also allow arbitrary runtime names to be used, like "Pat Fst Name" or whatever takes your fancy...then what the two sides have to agree on to get interoperability:

- data level interoperability - this requires information model agreement
- service level interoperability - this requires service model (communication viewpoint) model agreement
- knowledge description interoperability - this requires agreement on representation of archetypes, templates and the like by both sides.
- some agreement on actual archetypes & templates used - but this is not 100% necessary, since with the above 3, both sides can always communicate both data and understand the embedded domain structures


In this, I am assuming a fairly generic service interface that does not have things like "PatientFirstName" encoded into the functions (which I would argue is a mistake in any case).

he XML-RPC client and server have to agree on these terms
or no-go. This is where a published standard is a must.


The crucial point here is to realize the difference between a
"standard" and a "consensus" or "agreement" as you point
later in that mail.

Karsten

right - there will be some kind of contracts between health care facilities to use certain templates for example for transmitting referrals & discharge summaries around.

- thomas





Reply via email to