Artificial Intelligence in Medicine

This is an analysis of the characteristics of the principal existing medical expert systems describing the system developed by the author on first order logic. It attempts to set out in readily understood language the different methods of knowledge representation used in different systems with medical applications. The artificial intelligence approach seeks to bypass certain limitations of the purely statistical methods by developing a more structured perception of how to cope with the problems of diagnosis or therapeutic management. The problems of knowledge representation, knowledge acquisition, methods of inference, possibilities for explanation and problems of man-machine interface are at the root of the acceptability of information systems by the non-computer worker. All these various problems are considered through examples of existing systems before describing the SPHINX system and the results obtained in various diagnostic or therapy aid applications.