Computer-aided causal diagnosis of ascites. Analysis of a prototype

摘要

No computer program based on ES techniques is capable of replacing the expert physician in an absolutely guaranteed way. Four main problems are present when trying to process a human reasoning task: The knowledge representation; The analysis of the represented knowledge; To make explicit the way which the represented knowledge will be processed in; The capacity of the computer system to evolve when new situations (such as new knowledge or new reasoning ways) are found. New trends in the field of computerized diagnosis covering the above considerations are being developed. Inductive learning constitutes, perhaps, the most considered technique, providing an adequate way to represent, analyze and process specific knowledge about some concepts in order to find general descriptions of them. Moreover, a learning system is characterized by its evolutionary behaviour when new situations appear. A computer system based on new inductive learning techniques has been developed by the authors from the Computer Science Department of the Balearic Islands University, which has been applied as a prototype to the causal diagnosis of ascites. In the same way, information about this diseases has been provided by the authors from the Department of Gastroenterologia of the 'Son Dureta' Hospital.