That's a very interesting question, and one for which I don't really
have the answer. I don't know whether the non-relational (I prefer
not to call it "hierarchical" as I don't think that is really an
accurate description in this case) data model actually makes that
much of a difference with regard to data replication. I can see
arguments on either side. In my case, I lean a bit more to the "help"
rather than "hinderance" side, but consider the question open. with
regard to replication, there are a lot of small things, such as use
of RAID and SAN style Alpha clusters, but there seem to me to be two
(three?) factors here on the "macro" scale. For starters, there is
the Master Patient Index (MPI) which is used to maintain a unique
identifier for each patient (veteran) in the system. This is
important because it is the glue to that holds together information
stored on multiple systems. A second factor is the extensive use of
(HL7) messaging to facilitate exchange of information across systems.
A third factor is that there are a number of data repositories fed by
operational systems. Finally (I suppose this is a fourth factor), the
system is not really centralized. Though patients do have unique
identifiers, no one system is used to store data for everyone. The
operational databases at a facility store information for the
patients seen at that facility.
Unfortunately, messaging interfaces tend to be hand-crafted and
rather labor intensive to build. This is one issue I hoped to address
with my temporarily (I hope) suspended project that I initially named
HL7 Data Binding, though I later referred to it as either an
interface compiler or the generic query. Mostly, I've built pieces of
what I would eventually like to develop as parts of individual
projects. Lately, I've been thinking in terms of developing it
outside of VistA since there seems to be relatively little interest
in the project at present, and many coding techniques common in VistA
(such as direct sets to Fileman globals) are a real impediment to
making a generic solution work, and perhaps why hand-crafted
solutions have remained so common. I do have some thoughts about how
to address this problem (and have passed them on to my management),
but they could perhaps be considered a bit radical (such as building
a Fileman aware MUMPS interpreter, perhaps as a layer, rather like
VPE.) These days, I'm thinking of compilation as a much more
realistic solution.
===
Gregory Woodhouse
[EMAIL PROTECTED]
"Before one gets the right answer, one must ask the right question."
-- S. Barry Cooper
On Sep 10, 2005, at 12:09 AM, Mark Street wrote:
After reading the article on the VA getting the health records for all
veterens in New Orleans in 100 hours it got me thinking about the
underlying
infrastructure that allows the VA to deliver the goods in such a short
period.
http://govhealthit.com/article90691-09-08-05-Web&newsletter=yes
With some smaller clinics looking at Applications Service Providers
to provide
them with more bang for the buck and economy of scale for their
EMR, I was
wondering how VistA hierarchal location database/replication
architecture
compares and functions.
--
Mark Street, RHCE
http://www.oswizards.com
--
Key fingerprint = 3949 39E4 6317 7C3C 023E 2B1F 6FB3 06E7 D109 56C0
GPG key http://www.oswizards.com/pubkey.asc
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Hardhats-members mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/hardhats-members
