The accuracy of a Bayesian computer program for diagnosis and teaching in acute abdominal pain of childhood.

The use of a computer program for diagnosing acute abdominal pain in children using Bayes' theorem is described and the results of a one-year trial are analysed in order to qualify and quantify the accuracy of the system and to suggest some of the educational benefits that might be obtained. A means of feedback to medical students is described which can localise their weaknesses based on their experiences of real patients and could be used by educators to measure the effectiveness of their courses and the major areas of difficulty being encountered by the students. The computer has in this instance achieved a remarkable degree of accuracy in view of several factors acting against it. Such experience is very encouraging for further endeavour in this area.