基于决策树及神经网络的高血压病阴阳两虚证诊断模型的研究
|
摘要
目的:基于决策树及神经网络的方法,建立高血压病阴阳两虚证的诊断模型。方法:从古今医案及临床病例中收集高血压病病例,对所收集资料中患者的中医四诊信息进行归一化处理,并建立证候要素数据库,运用CHAID、CRT、QUEST及C5.0决策树算法和神经网络的方法提取高血压病阴阳两虚证的诊断规律,建立诊断模型。结果:采用CHAID、CRT、QUEST及C5.0决策树算法建立高血压病阴阳两虚证诊断模型,准确率分别为93.1%、91.5%、91.5%、96.03%,其中C5.0算法的诊断模型更优于其他3种;采用多层感知器(MLP,Multilayer Perceptron)、径向基函数(RBF,Radical Basis Function)建立高血压病阴阳两虚证诊断模型,前者训练样本正确百分比为95.9%,测试样本正确百分比为93.9%;后者训练样本正确百分比为99.2%,测试样本正确百分比为96.3%,径向基函数神经网络的诊断模型更优于多层感知器神经网络。结论:通过联合应用决策树及神经网络两种方法可见,腰膝酸软在高血压病阴阳两虚证的临床诊断中起决定性作用,同时结合中医四诊信息,可形成比较符合高血压病阴阳两虚证的诊断判别模型模式,为规范高血压病阴阳两虚证诊断标准提供依据。
Objective: To establish a diagnostic model of syndrome of deficiency of both Yin and Yang in hypertension based on the method of decision tree and neural network. Methods: The information of ancient and modern medical records and clinical cases of hypertension were collected. The collected four diagnostic methods information of TCM was normalized and the syndrome factor database was established and the C5.0, CRT, CHAID, QUEST decision tree methods and neural networks in SPSS 20.0 software were used for data analysis. The diagnostic rules were extracted from TCM clinical syndromes and finally the diagnostic models were established. Results: The diagnostic accuracy of CHAID, CRT, QUEST and C5.0 were 93.1%,91.5%,91.5% and 96.03%, respectively. The accuracy rate of C5.0 decision tree model was higher than that of the other three algorithms. Multi-layer perceptron neural network and radial basis functionneural network were used to analyze the original syndrome data. The training accuracy of the former was 95.9% and the test accuracy was 93.9%, while the training accuracy of the latter was 99.2% and the test accuracy was 96.3%. The diagnostic model of radial basis function neural network was better than that of the multilayer perceptron neural network. The results were consistent with the statistical requirements. The clinical diagnosis of Yin-Yang deficiency of hypertension was better guiding. Conclusion: Combination of decision tree and neural network data mining method shows weak waist plays a decisive role in the clinical diagnosis of Yin and Yang deficiency syndrome of hypertension, combining TCM four methods diagnostic information to form a combinatorial discriminant pattern that conforms Yin and Yang deficiency syndrome of hypertension to provide evidence for syndrome standardization study.
Objective: To establish a diagnostic model of syndrome of deficiency of both Yin and Yang in hypertension based on the method of decision tree and neural network. Methods: The information of ancient and modern medical records and clinical cases of hypertension were collected. The collected four diagnostic methods information of TCM was normalized and the syndrome factor database was established and the C5.0, CRT, CHAID, QUEST decision tree methods and neural networks in SPSS 20.0 software were used for data analysis. The diagnostic rules were extracted from TCM clinical syndromes and finally the diagnostic models were established. Results: The diagnostic accuracy of CHAID, CRT, QUEST and C5.0 were 93.1%,91.5%,91.5% and 96.03%, respectively. The accuracy rate of C5.0 decision tree model was higher than that of the other three algorithms. Multi-layer perceptron neural network and radial basis functionneural network were used to analyze the original syndrome data. The training accuracy of the former was 95.9% and the test accuracy was 93.9%, while the training accuracy of the latter was 99.2% and the test accuracy was 96.3%. The diagnostic model of radial basis function neural network was better than that of the multilayer perceptron neural network. The results were consistent with the statistical requirements. The clinical diagnosis of Yin-Yang deficiency of hypertension was better guiding. Conclusion: Combination of decision tree and neural network data mining method shows weak waist plays a decisive role in the clinical diagnosis of Yin and Yang deficiency syndrome of hypertension, combining TCM four methods diagnostic information to form a combinatorial discriminant pattern that conforms Yin and Yang deficiency syndrome of hypertension to provide evidence for syndrome standardization study.
引文
[1] 邱华丽,潘晓彦.高血压病防治研究现状[J].世界中西医结合杂志,2014,9(2):210-212.
[2] 牛子长,毛浩萍.原发性高血压病中医病因病机及证型分布规律的研究进展[J].辽宁中医杂志,2016,43(4):878-879.
[3] 庄华薇,戴建国,王忠山.高血压病的中医分期辨证论治[J].辽宁中医杂志,2015,42(4):885-887.
[4] 刘金涛,张腾.中医药防治高血压病作用机制及优势述评[J].中医杂志,2015,56(1):77-82.
[5] 陈潇雨,马利庄,胡义扬.基于决策树方法的慢性乙型肝炎中医证候分类[J].上海中医药大学学报,2013,27(1):40-44.
[6] 刘伟,丁长松,梁杨.数据挖掘技术在中医药领域研究中的应用[J].临床医药文献电子杂志,2016,3(35):7096-7097.
[7] 曾悦,张君.数据挖掘技术在中医药领域应用[J].中国中医药信息杂志,2012,19(3):99-100.
[8] 洪芳,何建成,曹雪滨.人工神经网络在中医证候研究中的应用现状与趋势[J].辽宁中医杂志,2013,40(1):13-15.
[9] 陈克龙,樊永平.数据挖掘中的分类算法及其在中医证候学中的应用[J].中华中医药杂志,2011,26(3):469-473.
[10] 《中国高血压防治指南》修订委员会.中国高血压防治指南[S].北京:人民卫生出版社,2010.
[11] 方药中,邓铁涛,李克光,等.实用中医内科学[M].上海:上海科学技术出版社,1985.
[12] 郑筱萸.中药新药临床研究指导原则[M].北京:中国医药科技出版社,2002:73.
[13] 王阶,熊兴江,刘巍.补肾法治疗高血压病[J].中国中药杂志,2013,38(9):1277-1279.
[14] 杨国棠,黄金花.高血压病中医证型分布规律的临床研究[J].中医临床研究,2013,5(12):34-35.
[15] 王清海,李桂明,李典鸿.高血压病中医证型分布规律的临床研究[J].新中医,2005(11):28-29.
[16] 谈学平,王承龙,骆吉鸿,等.原发性高血压病中医证型分布规律及其相关因素的临床研究[J].江苏中医药,2013,45(5):14-16.
[17] 张志斌,周春刚,陆曙.原发性高血压患者中医证型分布及其与生化指标的相关性[J].辽宁中医杂志,2010,37(6):969-971.
[18] 虞桂,何庆勇.7213例高血压病证候要素与应证组合分析[J].中华中医药杂志,2013,28(8):2284-2286.
[19] 杜峰,李运伦.高血压病阴阳两虚证证治规律的探讨[J].中国中医药现代远程教育,2016,14(9):48-50.
[20] 宋玉莲.温阳益肾汤治疗高血压病阴阳两虚证的临床疗效评价[J].医学理论与实践,2016,29(14):1881-1882.
[21] 张成秋,殷济清,辛青,等.金水宝胶囊联合氯沙坦钾干预高血压病阴阳两虚证患者早期肾损害的临床观察[J].中国中西医结合杂志,2013,33(6):731-735.
[22] 骆书彦,金宣菊.针灸结合中药治疗原发性高血压病(阴阳两虚型)30例[J].国医论坛,2004(5):17-18.
[23] Kass GV.An exploratory technique for investigating large quantities of categorical data[J].Applied Statistics,1980,29:119-127.
[24] Loh WY,Shih YS.Solit selection methods for classification trees[J].Statistica Sinica,1997,7:815-840.
[25] 吴超,张晓祥.数据挖掘在疾病诊断相关组项目中的应用[J].中国数字医学,2010,5(5):70-72.
[26] 帅健,李丽萍,陈业群.决策树模型与Logistic回归模型在伤害发生影响因素分析中的作用[J].中华疾病控制杂志,2015,19(2):185-189.
[27] 张玥娇,徐昕,代礼,等.CHAID模型在巩义市新生儿低出生体重影响因素筛选中的应用[J].现代预防医学,2012,39(1):18-22.
[28] 张超.分类树中CRT算法与判别分析的比较及其医学应用[J].数理医药学杂志,2008(2):139-141.
[29] 李强.创建决策树算法的比较研究——ID3,C4.5,C5.0算法的比较[J].甘肃科学学报,2006(4):84-87.
[30] 罗丽娟,段隆振,段文影,等.C5.0算法的改进及应用[J].南昌大学学报(工科版),2017,39(1):92-97.
[31] Cybenko G.Approximation by superpositions of a sigmoidal function[J].Mathematics of Control,Signals,and Systems(MC-SS),1989,2(4):303-314.
[32] 周慧,张尤赛,龚淼.基于RBF神经网络的医学图像分类算法研究[J].电子设计工程,2017,25(3):113-116,120.
[2] 牛子长,毛浩萍.原发性高血压病中医病因病机及证型分布规律的研究进展[J].辽宁中医杂志,2016,43(4):878-879.
[3] 庄华薇,戴建国,王忠山.高血压病的中医分期辨证论治[J].辽宁中医杂志,2015,42(4):885-887.
[4] 刘金涛,张腾.中医药防治高血压病作用机制及优势述评[J].中医杂志,2015,56(1):77-82.
[5] 陈潇雨,马利庄,胡义扬.基于决策树方法的慢性乙型肝炎中医证候分类[J].上海中医药大学学报,2013,27(1):40-44.
[6] 刘伟,丁长松,梁杨.数据挖掘技术在中医药领域研究中的应用[J].临床医药文献电子杂志,2016,3(35):7096-7097.
[7] 曾悦,张君.数据挖掘技术在中医药领域应用[J].中国中医药信息杂志,2012,19(3):99-100.
[8] 洪芳,何建成,曹雪滨.人工神经网络在中医证候研究中的应用现状与趋势[J].辽宁中医杂志,2013,40(1):13-15.
[9] 陈克龙,樊永平.数据挖掘中的分类算法及其在中医证候学中的应用[J].中华中医药杂志,2011,26(3):469-473.
[10] 《中国高血压防治指南》修订委员会.中国高血压防治指南[S].北京:人民卫生出版社,2010.
[11] 方药中,邓铁涛,李克光,等.实用中医内科学[M].上海:上海科学技术出版社,1985.
[12] 郑筱萸.中药新药临床研究指导原则[M].北京:中国医药科技出版社,2002:73.
[13] 王阶,熊兴江,刘巍.补肾法治疗高血压病[J].中国中药杂志,2013,38(9):1277-1279.
[14] 杨国棠,黄金花.高血压病中医证型分布规律的临床研究[J].中医临床研究,2013,5(12):34-35.
[15] 王清海,李桂明,李典鸿.高血压病中医证型分布规律的临床研究[J].新中医,2005(11):28-29.
[16] 谈学平,王承龙,骆吉鸿,等.原发性高血压病中医证型分布规律及其相关因素的临床研究[J].江苏中医药,2013,45(5):14-16.
[17] 张志斌,周春刚,陆曙.原发性高血压患者中医证型分布及其与生化指标的相关性[J].辽宁中医杂志,2010,37(6):969-971.
[18] 虞桂,何庆勇.7213例高血压病证候要素与应证组合分析[J].中华中医药杂志,2013,28(8):2284-2286.
[19] 杜峰,李运伦.高血压病阴阳两虚证证治规律的探讨[J].中国中医药现代远程教育,2016,14(9):48-50.
[20] 宋玉莲.温阳益肾汤治疗高血压病阴阳两虚证的临床疗效评价[J].医学理论与实践,2016,29(14):1881-1882.
[21] 张成秋,殷济清,辛青,等.金水宝胶囊联合氯沙坦钾干预高血压病阴阳两虚证患者早期肾损害的临床观察[J].中国中西医结合杂志,2013,33(6):731-735.
[22] 骆书彦,金宣菊.针灸结合中药治疗原发性高血压病(阴阳两虚型)30例[J].国医论坛,2004(5):17-18.
[23] Kass GV.An exploratory technique for investigating large quantities of categorical data[J].Applied Statistics,1980,29:119-127.
[24] Loh WY,Shih YS.Solit selection methods for classification trees[J].Statistica Sinica,1997,7:815-840.
[25] 吴超,张晓祥.数据挖掘在疾病诊断相关组项目中的应用[J].中国数字医学,2010,5(5):70-72.
[26] 帅健,李丽萍,陈业群.决策树模型与Logistic回归模型在伤害发生影响因素分析中的作用[J].中华疾病控制杂志,2015,19(2):185-189.
[27] 张玥娇,徐昕,代礼,等.CHAID模型在巩义市新生儿低出生体重影响因素筛选中的应用[J].现代预防医学,2012,39(1):18-22.
[28] 张超.分类树中CRT算法与判别分析的比较及其医学应用[J].数理医药学杂志,2008(2):139-141.
[29] 李强.创建决策树算法的比较研究——ID3,C4.5,C5.0算法的比较[J].甘肃科学学报,2006(4):84-87.
[30] 罗丽娟,段隆振,段文影,等.C5.0算法的改进及应用[J].南昌大学学报(工科版),2017,39(1):92-97.
[31] Cybenko G.Approximation by superpositions of a sigmoidal function[J].Mathematics of Control,Signals,and Systems(MC-SS),1989,2(4):303-314.
[32] 周慧,张尤赛,龚淼.基于RBF神经网络的医学图像分类算法研究[J].电子设计工程,2017,25(3):113-116,120.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |