Based On Bayesian Method To Research The Diagnosis Of Syndromes In TCM

Syndrome and clinical symptoms,are the clinical basis of diagnosis of Chinese medicine,with a variety of performance,the relationship complex,learning difficulties.Machine learning provides a new possibility,and Bayesian method is an important machine learning branch,that based on the Bayesian theorem's statistical classification method,with rigorous mathematical theory as a support.Through using the Bayesian method to analysis and identification the syndrome and clinical symptoms,easy to grasp the intrinsic link,for the study of TCM theory and clinical research gives useful tips.This dissertation focuses on two important algorithms about Bayesian method: naive Bayesian classification algorithm and Bayesian network.The main work of this dissertation as follows:1)Collect and summarize the TCM data from the "Chinese medicine diagnosis" in artificial way,applied to the Bayesian method.2)The naive Bayesian classification algorithm was applied to analysis of traditional Chinese medicine,and using the C# language to develop a form application program.Implements the classification and calculate the prediction accuracy,and add the Chinese characters' pinyin first letter prompts for input.3)Break through the conditional independence hypothesis of naive Bayesian classification,construct the network structure model of TCM diagnosis by using the concept of conditional information in information theory.The preliminary results of this dissertation are the use of naive Bayesian classification algorithm to develop TCM syndrome diagnosis classifier to achieve the classification of clinical symptoms of syndrome,classification accuracy rate can be more than 98%.The Bayesian network structure is constructed by the theory of graph theory and information theory.It is instructive to analyze the relationship between clinical symptoms and clinical syndrome.