Computing with words and machine learning in medical diagnostics

The paper deals with the mathematical simulation of empirical learning in medical diagnostics. Such a simulation essentially amounts to solving both the problem of how to acquire, represent, and objectify the smallest typical units of diagnostic knowledge and the problem of how to identify the subject-specific situations in which diagnostic learning may take place at all. A solution to the first question will be provided by using so called fuzzy probabilities as a model. The second problem is solved by adequately formalizing that kind of knowledge as it is found in standard sources like textbooks for instance. (C) 2001 Elsevier Science Inc. All rights reserved.