"Robert R. Hausam" <[EMAIL PROTECTED]> wrote:
>Alvin,
>
>I would say that you are probably correct, and that we shoud NOT reject
>VISTA out of hand due to its MUMPS (or now called M) implementation. It is
>a very complete and powerful system that is implemented and working on a
>production basis in I believe all of the VA sites (to one degree or
>another), and a number of other sites throughout the world.
I am aware of VistA users in Finland and Germany, as well as in the US and
several schools of Veterinary Medicine. I believe it is also the basis for
large medical systems deployed by the US Department of Defense and the US
Indian Health Service.
>I worked with
>the VA development office (IRMFO) as a fellow during the first 6 months
>that I was in Salt Lake, back in 1995. The Salt Lake office is where the
>development has been done for the clinical modules, including the Clinical
>Lexicon (their vocabulary component, which has some features worth looking
>at).
>
>While we are discussing MUMPS-based EMR systems, we should also mention
>COSTAR, which is another public domain system, developed at MGH originally
>for the Harvard Community Health Plan. It has also been widely and
>successfully used at a number of other sites.
I worked with COSTAR (Computer Stored Ambulatory Record) in 1981-1982. It
needed a rewrite then but it contained many good ideas and medical
knowledge. It was "frame" based with nearly all data stored in a
longitudinal patient record. We rejected it for use as a basis for *our HIS*
because it was too patient oriented, didn't have provision for enough
species of patients for our needs :-), and contained a deeply embedded
assumption that patients could be identified by social security numbers. I
don't know how it has progressed since then, but I believe a number of
people have developed GUI front-ends for it.
>Now for a few words about MUMPS. It is a very unstructured language (like
>BASIC, but probably worse?), but also very powerful, with built-in
>hierarchical database capability. It is particularly suited for string
>manipulation.
Any resemblance to BASIC is at best superficial. I think that MUMPS
is actually, best thought of, not as a computer language, but as a computing
environment conducive to the construction of highly scalable and resource
efficient multi-user information systems. This is reflected in the marketing
of InterSystems, the current owner of most of the commercial
implementations of MUMPS, for its flagship product, Cache. They refer
to Cache as "Post-Relational" and emphasize its advantages of performance
and flexibility over the relational databases together with its interfaces
to Java and SQL and the web. They never mention MUMPS, but that is its core.
The language itself is deceptively simple but rather different in syntax
from languages derived from C.
MUMPS' greatest strength, and the source of much its simplicity,
flexibility, scalability, and uniqueness, is the logical form of its shared
data storage (referred to simply as globals). MUMPS globals are multi-level
sorted associative arrays which are almost universally implemented on disk
with B+ trees (multi-way balanced trees).
>Due both to the nature of the language itself and also to
>how MUMPS programmers tend to use it, I think it would be difficult to port
>MUMPS routines to most other languages (i.e. the XECUTE command, if any of
>you are MUMPS programmers and know what I mean!). But MUMPS code and
>systems can be made to interoperate with other code modules and systems.
>In recent years a lot of work has gone on in the MUMPS community to improve
>interoperability and interfaces. I think these are fair statements. Art
>Smith or Jim Self may want to offer some additional comments or possibly
>corrections if they think I didn't get this quite right.
>
>My gut feeling is that we can learn a lot from the MUMPS-based efforts in
>VISTA and COSTAR, but we may not want to or be able to use any of their
>work directly in building the type of system for the future that most of us
>seem to be interested in.
VistA and COSTAR each embodies a tremendous amount of knowledge and
practical experience of medical information. They have each been installed
and proven effective in many hundreds of medical institutions.
To the best of my knowledge, there is nothing comparable outside of MUMPS.
I would be very surprised if the totality of all other open source medical
software has more than a small fraction of their functionality.
The main obstacle to the use of MUMPS software in open source projects is
the lack of an open source implementation of MUMPS itself. This is being
addressed with the GUM project and FreeM.
>Rob
>
>At 02:37 PM 11/19/99 -0500, you wrote:
>>> For information about Vista (including contact information), see
>>> http://www.hardhats.org
>>>
>>> -art
>>>
>>> Arthur B. Smith Voice: 573-884-4516
>>> W209 Vet. Med. Bldg. Fax: 573-884-4496
>>> 1600 E. Rollins Rd. Email: [EMAIL PROTECTED]
>>> Columbia, MO 65211 ICQ: 30211122
>>
>>
>>VistA. Could we perhaps be prematurely disregarding VistA as an option
>>just because it is based on MUMPS?
>>
>>If the "crucible of a model is in the implementation" (Wayne Wilson
>>right?), then VistA must be the most evidence-backed open source project
>>there
>>is...
>>
>>I am contacting Gregory Woodhouse to get his input...
>>
>>Alvin
>>
>>
>>
>
--------------------------------------
Jim Self [EMAIL PROTECTED]
VMTH Computer Services, UC Davis
Veterinary Medical Teaching Hospital
http://www.vmth.ucdavis.edu/us/jaself
