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

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.