A fuzzy-Bayesian model based on the superposition of states applied to the clinical reasoning support

This research was motivated by the need of a stochastic approach to inference processes on the complex knowledge that guides the medical decision. From this situation developed a mathematical model to Bayesian inference with inaccurate input variables and states of superposition characteristics. The domain of application and verification of the model is related to support clinical reasoning specialist to diagnose children and adolescents with metabolic risk. Bayesian techniques are used in the support for a diagnostic classification, however it is unprovided to infer about the variable which imprecise factor affects the accuracy of the results. Beyond the complexity that makes this scenario, a formal hybrid scheme that gathered Fuzzy Sets and Bayesian Techniques has been developed with the purpose of endorsing the specialist clinical reasoning with a casual and inaccuracy context relations. The model proposes an inference on overlapping states. From actual data records were done simulations applying the model and the results indicated that patients would gradually evolving into a high risk scenario while a Bayesian Expert System classic would be classified in less worrying levels. These results were also verified and confirmed by a team of medical experts consulted during this research. It is expected that the model can be attached to knowledge systems used by medical experts to guide their patients. It is believed to be applicable in other domains whose uncertainty is the biggest problem related to information or knowledge.