Knowledge engineering for very large decision-analytic medical models

Authors:

Marek J. Druzdzel
Decision Systems Laboratory
School of Information Sciences
and Intelligent Systems Program
University of Pittsburgh
e-mail: marek@sis.pitt.edu

Agnieszka Onisko
Bialystok University of Technology
Institute of Computer Science
Bialystok, 15-351, Poland
e-mail: aonisko@ii.pb.bialystok.pl
FAX: (085) 422-393

Daniel Schwartz
Center for Biomedical Informatics
University of Pittsburgh

John N. Dowling
Center for Biomedical Informatics
University of Pittsburgh

Hanna Wasyluk
The Medical Center of Postgraduate Education
and Institute of Biocybernetics
and Biomedical Engineering,
Polish Academy of Sciences
Warsaw, Marymoncka 99, Poland
e-mail: hwasyluk@cmkp.edu.pl

Abstract:

Graphical decision-analytic models, such as Bayesian networks, are powerful tools for modeling complex diagnostic problems, capable of encoding subjective expert knowledge and combining it with available data. Practical models built using this approach often reach the size of tens or even hundreds of variables. Constructing them requires practical skills that go beyond simple decision-analytic techniques. These skills are difficult to gain and there is almost no literature that would aid a modeler who is new to this approach. Drawing on our experiences in building large medical diagnostic models, we list typical problems encountered in model building and illustrate the knowledge engineering process with examples from our networks.

The full paper is available in PostScript (235KB) and PDF (108KB) formats.