The fuzzy knowledge-based system CLINAID: a correspondence between its relational and neural net structures

The paper describes in some detail the structure of inference in a generic family of multi-context knowledge based systems of CLINAID architectural type. The system is designed for knowledge-based decision making with risk and under uncertainty, operating in a multi-environmental situation and making decisions within a multiplicity of contexts. Because risk situations depending on a rare combination of risk factors as well as on the context in which the data are used appear not only in medicine, our approach has wider applicability than just the domain of medicine. The paper also pays attention to the translation of fuzzy relational structures into fuzzy neural network representation. The basic architecture consists of the following co-operating units (basic shell substratum): 1) diagnostic unit; 2) treatment recommendation unit; 3) patient clinical record unit; and 4) co-ordination and planning unit