Intelligent Decision Support in Medicine: back to Bayes?
Gitte Lindgaard (Carleton University, Canada)
Peter Egan (S4Potential, Canada)
Colin Jones (S4Potential, Canada)
Catherine Pyper (Carleton University, Canada)
Monique Frize (Carleton University, Canada)
Robin Walker (IWK Health Centre, Canada)
Craig Boutilier (University of Toronto, Canada)
Bowen Hui (University of Toronto, Canada)
Sheila Narasimhan (Carleton University, Canada)
Janette Folkens (Carleton University, Canada)
Bill Winogron (S4Potential, Canada)
Abstract: Decision Support Systems are proliferating rapidly in many areas of human endeavour including clinical medicine and psychology. While these are typically based on rule-based systems, decision trees, or Artificial Neural Networks, this paper argues that Bayes Theorem can be applied fruitfully to support expert decisions both in dynamically changing situations requiring the system progressively to adapt, and when this is not the case. One example of each of these two types is given. One provides diagnostic support for human decision makers; the other, an e-health mental intervention system provides decision rules enabling it to respond and provide the most appropriate training modules to input from clients with changing needs. The contributions of psychological research underlying both systems is summarized.
Keywords: Bayes' Theorem, Decision Support Systems (DSS), base rates, diagnostic error, e-health intervention, individuating information
Categories: H.1.2, H.4.2, I.2.1