Study And Application Of Support Vector Machines In Intelligence Medical Diagnosis System

Automated medical diagnosis is the direction of development of modem medical diagnosis. The thesis briefly introduces the study and application of intelligence medical diagnosis system, and presents an intelligence medical diagnosis system based on traditional Chinese medicine theory. This diagnosis system is composed of an embedded data acquisition equipment and a PC intelligence diagnosis platform. The data acquisition equipment collects the human impedance information; the PC diagnosis platform is used to set up the impedance database, preprocess the human impedance data, extract some feature parameters. Then use Support Vector Machine to train data and set up certain- model to predict uncertain swatch. The results of the analysis of hospital stomach trouble related illness cases, stroke cases and healthy cases show there are relations between human impedance feature parameters and different illness cases. Support vector machine can be used to classify the illness cases in diagnosis system, and given the support to the reasoning rules of expert system.The thesis respectively uses absolute value feature vectors and comparative value feature vectors to classify two types of hospital stomach trouble related illness cases, hospital stomach trouble related illness cases with healthy cases and stroke cases with healthy cases. The exactness results of the classification correspond to about 80%, 60% and 90%. Using comparative value feature vectors acquires better result of classification according to the analysis of result. RBF kernel behaves better compares to other kernel formats. According to the performance of Support Vector Machine in classification and the direction of development of intelligence medical diagnosis system it is worth further study when carrying out the research and development of the intelligence medical diagnosis system based on human impedance.