基于支持向量机的中风病中医证候分类  

Classification of TCM Syndrome Patterns of Stroke Based on SVM

作　　者：顾天宇 严壮志[1] 蒋皆恢[1] GU Tianyu;YAN Zhuangzhi;JIANG Jiehui(School of Communication and Information Engineering,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《中医药信息》2021年第9期1-3,共3页Information on Traditional Chinese Medicine

基　　金：国家重点研发计划项目(2018YFC1707704)。

摘　　要：目的:研究基于机器学习的中风病中医证候分类模型,推进中医证候分类的数字化、自动化与智能化。方法:利用支持向量机(SVM)、BP神经网络与梯度提升决策树(GBDT),以年龄、性别、身高、体质量与临床中风病常见的症状,包括中风主要症状、舌象、脉象数据作为输入,以中风临床常见证候--气虚血瘀证的判断作为输出,建立中风病中医证候分类模型并进行比较。结果:基于支持向量机建立的中风病中医证候分类模型准确率为86%,BP神经网络模型准确率为81%,梯度提升决策树模型准确率为84%。结论:相较于其他分类器模型,基于支持向量机建立的模型在中风病中医证候分类上更具优势。Objectives:To study the syndrome classification model of stroke in TCM based on support vector machine(SVM),and to promote the digitization,automation and intellectualization of its TCM syndrome classification.Methods:SVM,BP neural network and gradient boosting decision tree were used,with age,gender,height,body mass and common clinical symptoms,including main symptoms,tongue and pulse manifestations as input,as well as with the classification of common stroke syndrome of qi deficiency and blood stasis as output to establish the stroke TCM syndrome diagnosis models for the comparison.Results:The accuracy rate of the stroke TCM syndrome classification models was 86%based on SVM,the accuracy rate of the models was 81%based on BP neural network,and the accuracy rate of the models was 84%based on GBDT.Conclusion:Compared to other classified models,the stroke TCM syndrome classification models based on SVM perform better in the classification of TCM syndromes of stroke.

关 键 词：支持向量机 中风 证候分类 

分 类 号：R259[医药卫生—中西医结合]