New Approach to Treat Uncertainty in Diagnosing Cardiovascular Disease by Using Bayesian Theorem

A new approach to define and assign statistical parameters to Bayesian inference nodes derived from fuzzy logic technology is proposed. First to develop an intelligent medical diagnostic system, the individual membership function can be pre-defined by matching separately the adapted high-order polynomial, S-type or quasi-Gaussian function with plot of collected clinical diagnostic data. Consequently the coefficients in the defined membership function formulas are fixed which in the diagnostic process can be used to define membership grades vs. recorded symptoms dynamically and individually. Based on symptom-mapped membership grades, statistical parameters can be further defined and assigned to each relevant node in inference nets. The simplicity and adaptability of the proposed methodology is demonstrated and tested by applying it in diagnosing 5 most common and important cardiovascular diseases, through constructed hierarchical Bayesian fuzzy inference nets. The defined statistical parameters are used in calculating propagation of probabilities using Bayesian theorem to solve refractory uncertainty and deduce the disease(s). Key-Words: statistical parameters; uncertainty; membership function; Bayesian inference nets; propagation of probability, diagnosis of cardiovascular disease