基于舌脉象数据的决策树算法的非小细胞肺癌证候分类方法  

Classification of Syndromes of Non-small Cell Lung Cancer Based on Decision Tree Algorithm Based on Tongue Data and Pulse Data

作　　者：石玉琳 刘嘉懿 胡晓娟[3] 龚亚斌[4] 刘苓霜[5] 许家佗[2] Shi Yulin;Liu Jiayi;Hu Xiaojuan;Gong Yabin;Liu Lingshuang;Xu Jiatuo(Experiment Center For Teaching&Learning,Shanghai University of Traditional Chinese Medicine,Shanghai 201203,China;Basic Medical College,Shanghai University of Traditional Chinese Medicine,Shanghai,China;Shanghai Innovation Center of TCM Health Service,Shanghai University of Traditional Chinese Medicine,Shanghai 201203,China;Yueyang Hospital of Integrated Traditional Chinese and Western Medicine,Shanghai University of Traditional Chinese Medicine,Shanghai 200437,China;Longhua Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,Shanghai 200032,China)

机构地区：[1]上海中医药大学教学实验中心,上海201203 [2]上海中医药大学基础医学院,上海201203 [3]上海中医药大学中医健康服务协同创新中心,上海201203 [4]上海中医药大学附属岳阳中西医结合医院,上海200437 [5]上海中医药大学附属龙华医院,上海200032

出　　处：《世界科学技术-中医药现代化》2022年第7期2766-2775,共10页Modernization of Traditional Chinese Medicine and Materia Medica-World Science and Technology

基　　金：国家科学技术部国家重点研发计划中医药现代化研究重点专项(2017YFC1703300):中医智能舌诊系统研发,负责人:许家佗;上海市科学技术委员会启明星培育项目(22YF1448900):基于舌脉诊数据的NSCLC患病风险评估模型研究,负责人:石玉琳;上海市教育委员会预算内项目(2021LK029):基于中西医数据融合的肺结节与肺癌诊断分类模型研究,负责人:石玉琳

摘　　要：目的探究基于客观舌脉象数据的C5.0决策树算法的非小细胞肺癌典型证候分类方法的可行性,获取证候相关的客观舌脉象特征组合。方法以337例非小细胞肺癌患者的气虚证与阴虚证客观化舌象数据及脉象数据为研究对象,采用特征属性筛选及C5.0决策树算法数据挖掘方法,构建基于客观舌脉象数据的非小细胞肺癌气虚证与阴虚证的证候分类模型。结果基于脉象数据的模型分类准确率高于舌象数据,且基于单纯舌象数据或脉象数据的证候分类模型分类准确率低于基于舌脉联合的数据分类模型,舌脉联合数据的分类模型树深为5,从中筛选出了气虚证与阴虚证分类相关的7项重要属性及7条分类规则,模型分类准确率为80.37%。结论现代化舌脉象数据可以为非小细胞肺癌证候分类提供新视角,基于特征选择与C5.0决策树算法的非小细胞肺癌证候分类方法是可行的。Objective To explore the feasibility of the classification method of typical syndromes of Non-Small Cell Lung Cancer(NSCLC)based on C5.0 decision tree algorithm based on objective tongue data and pulse data,and to obtain the objective combination of tongue and pulse features related to the syndrome.Methods Tongue data and pulse data of 337 NSCLC patients with Qi-deficiency syndrome and Yin-deficiency syndrome were taken as the research objects,to construct the syndrome classification model of Qi-deficiency syndrome and Yin-deficiency syndrome of NSCLC based on objective tongue data and pulse data by using feature selection and C5.0 decision tree algorithm.Results The classification accuracy of model based on pulse data was higher than that of tongue data,and the classification accuracy of model based on tongue data or data was lower than that of integrated tongue data and pulse data.The classification model tree depth of the integrated tongue data and pulse data was 5,from which 7 important attributes and 7 classification rules related to the classification of Qi-deficiency syndrome and Yin-deficiency syndrome were screened out,and classification accuracy of the model was 80.37%.Conclusion The modern tongue data and pulse data could provide a new perspective for the syndrome classification of NSCLC,and the classification method based on feature selection and C5.0 decision tree algorithm was feasible.

关 键 词：非小细胞肺癌 证候 决策树 舌诊 脉诊 

分 类 号：R273[医药卫生—中西医结合] TP181[医药卫生—中医肿瘤科]