Computer based medical diagnosis

In order to provide an overview of computer based medical diagnosis the author sets out the major theoretical commitments involved. First he presents the AI problem solving paradigm, then the use of decision trees to map symptoms into diseases, next the need to apply Bayes theorem since symptoms are noisy and unreliable and the difficulties with doing so. The role of hypothesis and test and weak model of the causal processes is discussed (stronger models are difficult to use). He then notes the success of second generation expert systems in providing explicit representation of strategic reasoning and a structural representation of domain knowledge. Finally he concludes by describing the unifying capability of the second generation expert systems