ABSTRACT
This chapter discusses the use of causal Bayesian networks, and other causal networks, in modelling situations, which arise in medicine. It argues that Judea Pearl's suggested way of linking causality and probability is essentially a generalization of Reichenbach's approach. In 1980s Pearl introduced what he called "Bayesian networks". These can be thought of as generalizations of Reichenbach's conjunctive forks to complicated networks. In 2000, Pearl published a book on causality. In this work, his focus had shifted somewhat away from AI towards discussion of causal models in econometrics and epidemiology. In his 2011, Pearl gives an exposition of what he calls "the structural theory of causation". The main obstacle is of course what Pearl calls "probability raising" trap, or "the idea that causes raise the probability of their effects". Multi-causal forks are very simple causal models, which apply in a straight-forward way to well-known examples of the use of indeterministic causality in medicine, such as causal factors of heart disease.
