Application Of Support Vector Machine In TCM Syndrome Classification

The classification of TCM syndromes is the bridge between clinical medicine and TCM theory.The traditional classification of TCM syndromes is the type of syndrome that is diagnosed by the doctor through the combination of the signs of the patient and the clinical experience.In this process,not only the professional knowledge background is needed,but also the judgment of syndrome type depends more on the doctor's prior experience.In addition,the data dimension of TCM syndrome classification is large and the value density is low,so it is difficult to dig out the value of it in artificial way,which hinders the development of TCM syndrome classification.While machine learning as a kind of new technology,through the construction of learning model to predict unknown samples from known samples,in many machine learning algorithms,the algorithm of support vector machine with its unique learning ability and generalization ability is widely used,provides reliable technical support for the research of Chinese medicine syndrome classification.The main content of this article is to introduce how to use support vector machine algorithm to realize the classification of 125 kinds of TCM syndromes contained in the six Dialectical Categories of TCM syndromes.Through data preprocessing,125 kinds of TCM syndromes and 1084 kinds of different clinical symptoms of encoding and the distinction between the primary disease of each TCM syndrome and secondary disease,and select the main auxiliary disorder number is combined with the corresponding syndrome 1-7 combined form contains 960,590 records of training samples,then according to the dialectical type and size of the sample is divided into 10 small training samples.Then,we select two samples of syndromes,use the algorithm to implement the two classification,and can use the saved training model to predict the sample correctly.On this basis,1085 supervised multi class training samples are constructed,and the algorithm is used to predict the syndrome classification.By selecting different kernel functions and adjusting the parameters,the accuracy of the model is from the initial 34.8% to the final 98.9%,thus the classification of TCM syndromes is realized more accurately.