Computerized Pulse Analysis And Diagnosis Based On Bayesian Networks

Pulse Diagnosis is widely used as an important diagnostic method in Traditional Chinese Medicine(TCM). However, there are still several inevitable limitations that greatly obstruct the application of traditional pulse diagnostic in clinical medicine. Firstly, pulse states are described as a kind of special feeling on the fingertips of the examiner, which can't be well understood in western medicine. Secondly, pulse diagnosis requires lots of experiences and the interpretation is subjective, depending on the practitioner. Therefore, the diagnostic results may be unreliable and inconsistent. In this paper, we propose a novel computerized pulse inspection method aiming to address these problems. First of all, several key features are extracted from the pulse signals. Then, Bayesian networks are employed to model the relationships between these features as well as classify different pulse states. Furthermore, we set up a new Tree-Augmented Na?ve Bayesian Classifier model based on fuzzy partition, which is proved effective in dealing with both discrete and continuous variables.