Fuzzy Bayesian Conditional Probability Model and its Application in Differential Diagnosis of Non-Toxic Thyropathy

Abstract

Based on the Fuzzy mathematical principle, an all- propose fuzzy Bayesian conditional probability model (FBCPM) on differential diagnosis of non-toxic thyropathy is established. In this paper, the species of diseases consist in the state space, the information space is composed of the symptoms of diseases, attaching function is created based on the clinic experiences of the clinists and statistical principle, quantum standard of each symptom is determined under condition that medical experts take part in, and the posterior probability that the symptoms of the sufferer have come forth is defined. After the relative information, such as symptom signs of the sufferer, prior and conditional probabilities of each disease, is obtained from the case histories of the cases, attaching function values of all symptoms and posterior probability of each case can be calculated. The disease corresponding to the maximum of posterior probability is just diagnosed results. Finally, the performance of the FBCPM is tested using previous cases. The results has shown that accurate rate of the FBCPM is greater than that of the clinic medical experts via diagnosing the history cases.