Society for Medical Decision Making

Description and Objectives:

Probabilistic graphical models, developed in the field of artificial intelligence, offer important advantages over traditional methods for medical diagnosis and decision analysis: 

1. Bayesian networks are much more accurate than the naive Bayes method.

2. Influence diagrams and decision analysis networks are equivalent to decision trees containing thousands of branches. They can perform cost-effectiveness analysis on problems involving several decisions.

3. Markov influence diagrams can solve problems much more easily than when using spreadsheets, Markov decision trees, or a programming language such as R or C++. They can model various patient characteristics without multiplying the number of states; in particular, they can represent the patient's history without using tunnel states.

The purpose of this course is to introduce probabilistic graphical models and explain how to apply them to medical decision analysis using OpenMarkov, an open-source tool. In particular, participants will learn how to: 

• build Bayesian networks for medical diagnosis and compare them with naive Bayes models;

• build decision analysis networks for unicriterion and cost-effectiveness analysis and compare them with decision trees;

• build Markov influence diagrams for cost-effectiveness analysis and compare them with spreadsheets, Markov decision trees, and discrete event simulation.

The course will also present other temporal models, such as Markov decision processes (fully and partially observable) and dynamic limited-memory influence diagrams.

Register Now | Return to Short Course Listings