Introduction and Invitation
Mon, 29 Apr 96 20:53:01 +0500

Duane Steward, Introduction

I received my Doctor of Veterinary Medicine degree from the University of
Florida in 1980. Do not mistake me, however, for a "Gator" because both my
bachelors degree in biology (1976) and my MSIE(1994) were conferred by
Florida State University (Seminoles).

I practiced veterinary medicine in Fla. over a fourteen year period in the
Keys, Ocala, and Tallahassee. The longest period was the most recent
eleven years as sole proprietor of a housecall practice. Although my
formal training in psychology is limited to one undergraduate course in
FSU's psychology department and one animal behavior course in the biology
department, an understanding of behavior has seemed to be an essential
element of successful practice in veterinary medicine. The animal patient
needs to be understood to successfully treat the needs of one who does not
"understand", at least at the outset, what you are trying to do. The client
(ie, patient owner) needs to be understood in order to treat successfully
as well. Additional help is often required of the client to achieve the
desired effect for the patient which requires an additional layer of
understanding. Finally, the relationship between the pet and owner is often
complicated with many demands for understanding. The psychological
processes involved are most certainly to be ignored only at one's peril.

It is natural then that I became involved as a consultant to the Florida
State Area Agency on Aging's program to promote pet-facilitated therapy. I
was involved from its inception as a veterinarian who monitored the growing
body of literature on the human-animal bond and participated in the
screening of animal candidates for use in visitation and institutional

My success in practice created a threat of "burnout" that, combined with a
difficulty in hiring staff with matched long term commitment, resulted in a
professional extension outside veterinary medicine. It began with a
conviction that there must be a better way to solve problems and that other
professions, engineering for instance, could offer insight. I found myself
pursuing a Masters degree in Industrial Engineering at the FAMU/FSU College
of Engineering while still seeing my housecall patients. This was made
possible with cutting edge technologies that included cellular telephone,
laptop computers and electronic bulletin board consulting communication.
While learning the basics of Operations Research, Stocastics and
Optimization, I was also learning of a new domain in medicine which had an
infantile counterpart in veterinary medicine: "Informatics". I came to
realize the overlay between industrial engineering and veterinary or
medical informatics.

Opportunities arose that beseeched my leaving practice. Having completed
the Masters degree, I am now a fellow in the National Library of Medicine's
Research Training Program in Medical Informatics. This grants me the
opportunity to be a candidate for a PhD in computer science at M.I.T. as a
member of the Harvard/MIT Health Sciences and Technology group. Leading
projects of my actual research group include the Heart Disease [diagnosis]
Program, World Wide Web - Electronic Medical Record System, and Guardian
Angel. The first two have explanatory titles. The Guardian Angel is a
developing concept of a patient located medical record repository networked
with health care provider information systems allowing 1) single target
dependability for complete health records, 2) tighter and intelligently
mediated feedback loops for intensive medical case management and 3) record
based gateways to health care information infrastructure to facilitate
education and learning about personal health specifics,

My personal research interest has its seed in the work of Wyllis Bandler,
Evelyn Stiller, and Vasco Mancini at FSU who sought to provide urban
planners with the "expertise" of common residents using the results of
repertory grid interviews. I wish to capture the "expertise" of an
individual patient in the health care process and provide it to his/her
physician in a suitable form for artificial intelligence applications in
diagnostic and treatment aides. Personal constructivist perspectives
provide a well featured paradigm for knowledge representation. The
elaboration of attributes as bipolar constructs rather than single
characteristics is a promising restriction of feature space. The presence
of hierarchic structure allows the propagation of constraints, hence
further restriction. Such restrictions have been the key to many first
successes in artificial intelligence. The structure of relationships
between constructs potentiates the capacity to exploit the feature space to
efficiently reason with machine algorithms. My hope then, is to formalize a
knowledge representation which can be programmatically exploited to
facilitate understanding between physicians and patients (or animal
owner/custodians). I hope to be able to discern the objects of an
individual's world view by repertory methodology and to capture the
relation between objects such that analysis can facilitate communication
and validation.

Just as Kellians hold that psychological processing results in merely an
image of the real world as "interpreted" by the individual, artificial
intelligence has recognized that all knowledge representation schemes are
incomplete, leaving out some degree of detail from the physical reality of
experience. I do not propose to engineer an architecture that will capture
complete individual perspectives. However, I do wish to exploit the feature
space of the healthiness/unhealthiness construct to pinpoint the patient's
perspective in medical decision making. I propose that triad presentation
methodology can be employed over the question of "how are two of them
healthier or less healthy that the third?" The result, I maintain, should
allow us to identify the objects of the patient's view on what healthiness
is and analysis thereof enable us to discover objective relationships
between these objects. Consequent integration of such objects and relations
in the erection of medical decision models and specific outcome options
suggest improved compliance and more effective utilization of resources as
well as reduced stress on the health care providership stemming from a
clearer understanding. I additionally hope to find the ability to classify
people according to their constructus substructures if clustering can be
observed to facilitate stochastic reasoning.

I have piloted this interest by conducting my first efforts at classic
repertory grid interviews of subway passengers initiating with just exactly
the question I proposed above. I have found later when I would attempt to
reason about the inter-construct structure, I lacked the ability to soundly
address possible transformations born, for example, of transitivity or
reciprocity. To disambiguate, I propose to formalize the structure of the
interview enough to support the intentions I have for the outcome of the
interviews. As a result of my pilot experience, I have developed an
interest specifically in the patient's perspective on the following list of
properties regarding the constructs they use to discriminate healthy from
unhealthy: implication, causality, classification of range element type,
controllability, activity, endocentric vs. exocentric, and intrinsic vs.
extrinsic character. This is my current experimental set of construct
features by which I propose to reduce the feature space for tractability
and exploit for expressiveness and utility in understanding medical
decision making behavior. I am currently attempting to use it to write an
expert system that assists the interviewer in repertory grid triad
presentation. I could use some expert help in critiqueing the rules of this
expert system as they arise.

I would like to invite any and all interested parties to browse my web page
for more details. Of particular interest should be the link to "Patient
Desires for Health".

Forewarned, I have indeed found a scarcity of Kellyians in Cambridge, MA.
This listserv makes that less consequential if I can find willing
colleagues to critique and suggest. I am NOT a psychologist and invite
collaboration. I think my work would be far more credible if I were to find
a suitable collaborator. My role in this work would be best kept to that of
knowledge engineer with some experience in medical pragmatics. If you are
especially triggered by my diatribe here, you are invited to email me
directly. The web site and contact information is present in my signature.


Duane Steward, D.V.M., M.S.I.E., Fellow A.A.V.I.
Fellow in Medical Informatics
Clinical Decision Making Group; Laboratory for Computer Science; M.I.T.
NE43-415 545 Technology Square Cambridge, MA., 02139 URL:
(617) 253-3533 Group Office: 253-5860 Fax: (617) 258-8682