The Center for Biomedical Informatics
State University of Campinas, Brazil

Research Abstracts

The process of creating computer-usable knowledge bases (KB) is complex, lengthy and still poorly understood by most health care professionals. There is a clear need for intermediate courses which teach knowledge engineering (KE) concepts and methods for these professionals. Understanding better its steps and requirements, they will be able to interact more efficiently with expert system builders. We describe here a 4-year experience with a practical course on Knowledge Engineering and Artificial Intelligence applications to Medicine. It is a one-semester, 64-hour course ministered at graduate level at the Biomedical Engineering program. The participants are a mix of physicians, human biologists, nurses and biomedical engineers pursuing a Master's degree either in Medicine, Physiology or Biomedical Engineering. The course contents addresses theoretically and practically the main methods used to develop decision-making tools: algorithm-based systems, binary and quantitative pattern classification, multivariate statistics, Bayesian systems, logical and rule-based programming and connectionist systems. For each method, the students receive a small, easy-to-use microcomputer program which includes a worked-out example and several proposed exercises, based on real decision-making problems in Medicine. These 12 programs were developed by us, using a common user interface, for IBM/PC-compatible microcomputers, and are available in the public domain. An 8-hour introduction to PROLOG, complements the course. One KE group project using a rule-based expert system shell must be completed. Several decision trees in medical domains are provided to the students for this, or alternatively, they may choose their own problems. Additional seminars cover topics such as the nature of medical information and of decision-making in Medicine, the history and characteristics of KE in Medicine, and how some real-world expert systems work. Although limited in time, the course has been able to provide the students with a working knowledge on the scope and methods of computer-based KE in the Health Sciences and has achieved a good success. The extensive practical work with programs and examples related to medical problems has improved the students' motivation and understanding of the basic methodological approaches.

Published in:

Return to Home Page

Return to Abstracts Index

Last Updated: March 2, 1996

renato@sabbatini.com