[Aid to diagnosis and decision making in an acute abdominal pain syndrome].

Diagnostic and decision making performances of a Bayesian model have been compared with clinically performances in abdominal pain of acute onset. Diagnostic accuracy of computer (63.3 p. cent) was lower than diagnostic accuracy of clinicians (72.6 p. cent). A two-fold increase in the number of diagnoses similarly decreased both performances. When the computer came to use national data base (6 916 patients) instead of local data base (571 patients), diagnostic accuracy of low-prevalence diseases increased and diagnostic accuracy of high-prevalence diseases decreased. Decision making accuracy of computer and clinicians were identical.