Metapattern recognition in medical diagnosis

Medical diagnosis is a classification problem. Individual patients are assigned to one or more classes of illness on the basis of their symptoms. There are several complementary and competing approaches to computer based classification, including neural networks and expert systems. These approaches are discussed as examples of pattern recognition. Metapattern recognition (MPR) provides an environment in which the different pattern recognition methods may be combined to give an optimal result. The output from MPR is an optimised pattern recognition system and an estimation of its classification accuracy.<
>