摘要
目的建立基于机器学习的胃食管反流病(GERD)中医智能辨证模型。方法基于统一的中医证候量表,收集胃食管反流病符合肝胃郁热证、中虚气逆证临床病例共98个样本,2组证候各49个样本,并按照类别之间1∶1比例,全部数据的70%作为训练集,30%作为测试集。根据信息的"有、无"分别赋值"1、0",建立胃食管反流病中医临床信息数据库。应用支持向量机(SVM)、神经网络(NNs)和自动编码器(Autoencoder)分别构建GERD智能辨证模型,比较证候预测的准确性。结果98例GERD患者中,2组证候的体重指数(BMI)比较,差异有统计学意义(P<0.05)。而在幽门螺杆菌感染(Hp感染)情况、匹兹堡睡眠质量指数中差异无统计学意义(P>0.05)。在相同训练、测试样本数据下,支持向量机、神经网络和自动编码器+神经网络这三种算法对胃食管反流病两种证候的识别准确率分别为78.3%和79.2%和79.2%。结论NNs及Autoencoder降维基础上的NNs模型具有很好的诊断、预测能力,机器学习技术应用于GERD辨证模型的构建具有方法学上的可行性。
Objective To establish a TCM intelligent pattern differentiation model of Gastroesophageal Reflux Disease(GERD)based on Machine Learning.Methods According to TCM pattern differentiation standards,a total of 98 samples were collected from the clinical cases of GERD Gnsitent with the diagneses of stagnated heat in liver and stomach pattern,and spleen-hypofunction and reverse qi pattern.49 samples were collected for each pattern,forming two groups.Based on a 1:1 ratio between categories,70%of all data were used as the training set and 30%as the test set.According to the"presence"and"absence"of the information,the value"1"and"0"were respectively assigned to establish the TCM clinical information database of GERD.Then a pattern differentiation model of GERD was establishedwith the Support Vector Machine(SVM),Neural Networks(NNs)and Automatic Encoder(Autoencoder)respectively.The accuracy of pattern prediction of the three methods was compared.Results There was a statistically significant difference in BMI between the two groups(P<0.05).However,there was no significant difference in H.pylori infection or Pittsburgh sleep quality index(PSQI)(P>0.05).With the same training and test sample data,the recognition accuracy of those two pattern with SVM,NNs and Autoencoder+NNswere78.3%,79.2%and 79.2%respectively.Conclusion NNs and NNS+Autoencoder model seems to be effective in diagnosis and prediction of those two patterns.Methodologically,it might be feasible to establish a TCM intelligent pattern differentiation model of GERD based on machine learning.
作者
曹云
卢毅
陈建新
何莹
刘凯文
符欣
黄佳钦
丁霞
方俐晖
李志红
Cao Yun;Lu Yi;Chen Jianxin;He Ying;Liu Kai wen;Fu Xin;Huang Jiaqin;Ding xia;Fang Lihui;Li Zhihong(Dongzhimen Hospital,Beijing University of Chinese Medicine,Beijing 100700,China;The State Key Laboratory of Management and Control for Complex Systems,Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China;Beijing University of Chinese Medicine,Beijing 100029,China)
出处
《北京中医药大学学报》
CAS
CSCD
北大核心
2019年第10期869-874,共6页
Journal of Beijing University of Traditional Chinese Medicine
基金
国家重点研发计划资助项目(No.2018YFC1704100、No.2018YFC1704106)
吴阶平医学基金会临床科研专项资助项目(No.320.5750.17233)~~
关键词
支持向量机
神经网络
自动编码器
胃食管反流病
中医智能辨证模型
support vector machine(SVM)
neural networks
autoencoder
gastro-esophageal reflux disease(GERD)
TCM intelligentpattern differentiationmodel
