Diagnosis of nephropathia epidemica by adaptation through Lukasiewicz inference
Patrik Eklund, Jari Forsström · 1990
This paper describes a method how to arrive at a medical expert system (as a knowledge based system) to support physicians in classifying patients in diagnosis of Nephropathia epidemica (NE). We thereby present a link between Lukasiewicz inference and learning in neural nets, as a formal connection between uncertainty in logical implication and synaptic weights. The system presented uses clinical findings and laboratory investigations to arrive at predictions whether or not patients suffer from NE. Although we are willing to call our system a medical expert system, it could equally well be called a decision support system, this being more in spirit to what such a system really offers a physician.