At 11:47 PM 11/20/99 -0500, Alvin Marcelo wrote:
>Dave said:
>
> > I don't actually think people should have the same tables. This is too
>low
> > of a level
> > of commonality, in my opinion. The middleware approach says we will
> have a
> > mechanism
> > for specifying how information will be exchanged and requested including
> > how to define
> > new terms, translate terms, etc. People should be able to create new
> data
> > types, etc.
> > and still have things interoperate. Certainly things like ICD-9 data, of
> > course would be standard.
>
>Aha...now it is starting to make sense (to me!).
>
>I don't know if anyone else has noticed this, but the list is actually
>made up of two kinds of people: the basic (pure) researchers and the
>clinical (applied)
>researchers. The basic researchers (like the developers of GEHR and HL7
>3.0) intend to make a generalizable method of storing medical
>information that will be flexible enough to accommodate yet undetermined
>concepts.
>The clinical researchers (like John, Alex, Tim et al) are those whoput
>the models to test by actually implementing them.
We'll we've implemented a number of the CORBAmed specs in real models (as well
as participating in their creation). We typically are using HL7 data in
the systems, but
the systems are not constrained to handle only HL7 data.
>GERH and HL7 3.0 are the only models now on the block. CORBAmed it seems
>would be on a different (albeit very important) category ---- seamless
>interfacing. If our discussions fall through and we fail to come to a
>consensus, a commitment to a minimum of CORBAmed interfaces may be our
>final decision.
You can implement GEHR and HL73.0 with CORBAmed specifications. They are
not in opposition to those "standards". By doing so, you can have both of
them in your system, if you like
or you can have a GEHR system working with an HL73.0 system. The CORBAmed
models
can accommodate both.
>However, if we work fast, we can decide on a model soon and perhaps agree
>on a level as low and primitive as table design (since many of the
>projects are just beginning anyway).
Our project is 6 years old. The most difficult problem I've faced is
trying to
get someone's adhoc table design plugged into CORBAmed interfaces. This is
doable and the intention of a number of vendors, but is much easier if the
interfaces
are considered in the table design process.
>Chris (or Thomas Beale) and Gunther:
>
>Please respond categorically and if possible, provide a reference.
>
>1. Is it possible to extract a subset of your model enough to build a
>primary care record?
>
>2. If yes, can you provide us with this subset (please reference)?
>
>3. This question is relevant to me (but maybe to John Gage too):
>
>How do I then build my relational SQL tables from this subset? Or if
>someone prefers to use an object-oriented dbase, how can he do this too?
Why not create a system design rather than SQL tables?
Dave
>Will appreciate your responses..
>
>Thank you.
>
>Alvin
