website

Mar 28 2011

Rehashing MUMPS/Data Dictionary vs. Relational Model

Published by at 9:19 pm under VistA

I saw my name come up in an discussion thread questioning my understanding of relational databases and presumably, my rationale for the data dictionary-based model underlying FileMan and VistA..  Since I don’t respond to anonymous threads, I thought I’d post a few words here.

This is another never-ending story, now going 35 years.  It seems that there are these Mongolean hordes of people coming over the horizon, saying the same thing about treating medical informatics as just another transaction processing system.  They know banking, insurance, or retail, so therefore they must understand medical informatics as well.

I looked very seriously at the relational model, and rejected it because I thought it was too rigid for the expression of medical informatics information.  I made a “grand tour” of the leading medical informatics sites to look at what was working for them.  I read and spoke extensively with Chris Date http://en.wikipedia.org/wiki/Christopher_J._Date , Stanford CS prof Gio Wiederhold http://infolab.stanford.edu/people/gio.html (who was later to become the major professor of PhD dropout Sergy Brin), and Wharton professor Richard Hackathorn.  I presented papers at national conventions AFIPS and SCAMC, gave colloquia at Stanford, Harvard Medical School, Linkoping University in Sweden, Frankfurt University in Germany, and Chiba University in Japan.

My data dictionary work that found its way into DHCP was actually the third generation of a sequence of efforts that I had been working on.  The first was an OS/360 using Macro Assembler language, the second was in MIIS on a Data General Nova, (called LUMPS, for Loma Linda University MIIS Programming System).

Somehow, rehashing this issue a third of a century after my approach has been proven a success – seems like arguing about buggywhips and motor cars.
I’m appalled at how little forward-thinking architectural design work I see in Health IT…. everything I see is paving the cowpaths, pushing ever harder in the “integration crunch” mentality of the past 40 years.  Its a little like an organization that has lots of isolated intercoms, and can only look at integrating the intercoms rather than replacing them with a telephone system.

Where is the innovation beyond the current cowpaths?  Can we really be back to rehashing 35 year old designs?  Everyone talks about innovation, just like every clinic in America says that it is “patient centered” (which really means, “you come first after us”).  But I see precious little out there.

Why aren’t we looking at scale-space architectures or Space-Based architectures?  How about innovative representation of time?

Share

Comments Off

Comments are closed at this time.

Creative Commons License
Images by Tom Munnecke is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License.
Based on a work at munnecke.com.
Permissions beyond the scope of this license may be available at munnecke.com/license.