On a probabilistic set covering model for diagnosing psychiatric disorders.

Correct diagnosis and treatment of the patients attending a hospital is the foremost requirement if that hospital is to be a profitable proposition. F.T. DeDombal achieved 70% reduction in unwanted surgeries and almost 100% reduction in delayed surgeries in cases of acute abdomen through computer-aided diagnosis. Similarly, R. Boom et al proved an increase in the professional efficiency of internees in diagnosing cases of jaundice, acute abdominal pain & upper G.I. bleeding after they started using computer. N.G. Rao reported diagnostic accuracy of 94% for various types of anemias with the use of expert system developed by him. Johri & Guha showed an enhancement of performance of non-expert clinician in tackling psychiatric disorders from 64% to 86%. Many developed countries have developed data bases for studying the epidemiological patterns of various diseases for future health planning and combating strategy. Another advantage of computer based diagnosis is that it will provide uniformity for defining diseases, symptoms & signs which is very important for pursuing medical research. A very simple methodology is described in this paper which can provide data driven as well as heuristic driven approach to medical decision making. To start with, this approach has been used to analyse psychiatric disorders but can be implemented elsewhere with the same efficiency.