摘要
目的建立原发性肝癌肝郁脾虚证诊断模型,形成原发性肝癌肝郁脾虚证判别模式,挖掘其核心诊断属性,为进一步研究原发性肝癌标准化提供依据。方法搜集2014年6月1日-2019年6月1日湖南省中医药研究院附属医院肿瘤诊疗中心原发性肝癌住院患者的病症信息,进行规范,经2名主任医师进行二次辨证,建立原发性肝癌中医病症-证候数据库,运用CHAID(卡方自动交互检测)、QUEST(快速、无偏、高效统计树)、CART(分类回归树)、C5.0决策树算法及贝叶斯网络建立诊断模型。结果共纳入患者741例,包括肝郁脾虚、肝胆湿热、脾虚湿困、肝肾阴虚、肝热血瘀5个证型。测试样本验证结果显示,CHAID、QUEST、CART、C5.0决策树算法判别正确率分别为91.26%、90.86%、91.47%、92.67%,C5.0正确率略高于其他3种;贝叶斯网络分析显示,各病症存在一定关联,如肝区疼痛-脘腹胀满,脘腹胀满-纳呆厌食,倦怠乏力-纳呆厌食,肝区疼痛-纳呆厌食,脉细-脉弦细,脉弦-脉弦细,夜寐欠安-苔少,舌淡-舌胖,苔白-苔少,口干-口苦,双下肢浮肿-舌淡,苔白-脘腹胀满;在贡献度方面,排名前8位病症分别为脉弦细、纳呆厌食、口干、舌淡、倦怠乏力、肝区疼痛、口苦、脘腹胀满,与决策树算法结果基本吻合。结论决策树及贝叶斯网络均可从繁杂、无序的数据库中挖掘出原发性肝癌肝郁脾虚证核心诊断属性;脉弦细在肝郁脾虚证诊断中起决定性作用,结合肝区疼痛、舌淡、倦怠乏力、口干、口苦、纳呆厌食等病症信息,可形成比较符合肝郁脾虚证的判别模式,为原发性肝癌肝郁脾虚证提供较客观的诊断依据。
Objective To establish a diagnosis model of liver stagnation and spleen deficiency syndrome of primary liver cancer;To form an identification mode for liver stagnation and spleen deficiency syndrome of primary liver cancer;To mine its core diagnostic attributes;To provide a basis for further research on the standardization of primary liver cancer.Methods The disease information of inpatients diagnosed with primary liver cancer in Tumor Diagnosis and Ttreatment Center of Affiliated Hospital of Hunan Institute of Traditional Chinese Medicine from 1st June 2014 to 1st June 2019 was collected,and the information was standardized,unified,and received the second syndrome differentiation by 2 chief physicians.A database of TCM syndromes-symptoms of primary liver cancer was established.A diagnosis model was established by using decision tree of CHAID,QUEST,CART,C5.0 algorithm and Bayesian network.Results Totally 741 patients were involved,including 5 syndromes of liver depression and spleen deficiency,liver and gallbladder dampness-heat,spleen deficiency and dampness,liver and kidney yin deficiency,liver heat and blood stasis.The results of test sample verification showed that the correct rates of CHAID,QUEST,CART,C5.0 decision tree algorithm were 91.26%,90.86%,91.47%,and 92.67%,respectively,and the correct rate of C5.0 was slightly higher than that of the other three types.The results of Bayesian network analysis showed that there was a certain correlation between the symptoms,such as liver pain-epigastric distension,epigastric distensionanorexia,burnout and fatigue-anorexia,liver pain-anorexia,pulse thin-pulse string thin,pulse string-pulse string thin,sleeplessness-less moss,tongue light-tongue fat,moss white-moss less,mouth dry-mouth bitter,both lower extremities edema-tongue light,moss white-epigastric distension,etc.In terms of contribution,the top 8 disease symptoms were pulse string thin,anorexia,mouth dry,tongue light,fatigue,liver pain,mouth bitter,and epigastric distension,and the results basically agreed with
作者
张振
田雪飞
郜文辉
何凤姣
邓天好
宋晓燕
郑飘
黄振
ZHANG Zhen;TIAN Xuefei;GAO Wenhui;HE Fengjiao;DENG Tianhao;SONG Xiaoyan;ZHENG Piao;HUANG Zhen(School of Integrated Chinese and Western Medicine,Hunan University of Chinese Medicine,Changsha 410208,China;Tumor Diagnosis and Ttreatment Center of Affiliated Hospital of Hunan Institute of Traditional Chinese Medicine,Changsha 410008,China;School of Mathematics and Metrology,Hunan University,Changsha 410208,China)
出处
《中国中医药信息杂志》
CAS
CSCD
2020年第9期115-120,共6页
Chinese Journal of Information on Traditional Chinese Medicine
基金
国家自然科学基金(81603603、81473617)
湖南省教育厅开放平台基金(16K066)
湖南省科技计划(2017SK50310)。
关键词
原发性肝癌
肝郁脾虚证
决策树
贝叶斯网络
primary liver cancer
liver depression and spleen deficiency syndrome
decision tree
Bayesian network
