基于Logistic回归分析的中风常用中成药适宜人群识别的示范性研究  

Logistic Regression on Recognition of Clinical Features of Proper Population of Chinese Patent Medicine Commonly Used in Stroke

作　　者：罗文舒[1] 胡湘[1] 于海波[1] 张金文 黄美玲 崔梦媛 Luo Wenshu;Hu Xiang;Yu Haibo;Zhang Jinwen;Huang Meiling;Cui Mengyuan(Shenzhen Traditional Chinese Medicine Hospital,Shenzhen 518033,China;The Fourth Clinical Medical School of Guangzhou University of Chinese Medicine,Shenzhen 518033,China;Guangzhou University of Chinese Medicine ,Guangzhou 510006 ,China)

机构地区：[1]深圳市中医院,广东深圳518033 [2]广州中医药大学第四临床医学院,广东深圳518033 [3]广州中医药大学,广东广州510006

出　　处：《亚太传统医药》2019年第1期134-136,共3页Asia-Pacific Traditional Medicine

基　　金：深圳市科技计划项目(JCYJ20140408153413105);深圳市科技研发资金条件与平台建设计划重点实验室项目(CXB201111250113A)

摘　　要：目的:应用Logistic回归分析研究中风常用中成药适宜人群的临床特征。方法:查询电子处方集系统,筛选应用量最大的前3种中成药,对常用中成药适宜人群,进行望、闻、问、切直接采集数据,录入数据进行统计分析,利用Logistic回归分析建立数学模型,分析模型对临床特征分类的预测能力。结果:筛选出应用量最大的前3种中成药为灯盏生脉胶囊、银杏叶滴丸、强力天麻杜仲胶囊,共纳入203例受试者,利用Logistic回归分析构建了数学模型,对3种中成药组测试集进行预测,判别正确率分别为81.82%、77.78%和78.13%,总判别正确率为79.21%。结论:利用Logistic回归分析构建中成药适宜人群的数学模型,对3种中成药适宜人群临床特征有良好的识别效果。Objective:To investigate the recognition of clinical features of proper population of Chinese patent medicine commonly used in stroke by Logistic regression.Methods:The first 3 Chinese patent medicines with the largest number of applications have been screened by inquiring of electronic medical records.Patients’ data of the proper population for 3 Chinese patent medicines is collected by Inspection,Auscultation-Olfaction,Inquiry and Pulse-taking-Palpation which are four diagnostic methods of Chinese medicine.The logistic regression is performed to construct the mathematic model for the proper population of Chinese patent medicine commonly used in stroke.Results:The first 3 Chinese patent medicines with the largest number of applications are Erigeron Shengmai Capsule,Ginkgo biloba drop pills and Powerful Gastrodia Eucommia Capsule.The mathematic model is constructed by Logistic regression for 203 subjects.The recognition accuracy for 3 groups of Chinese patent medicines are 81.82%,77.78% and 78.13%respectively.Total recognition accuracy is 79.21%.Conclusion:The recognition accuracy of the mathematic model for proper population of Chinese patent medicine constructed by logistic regression is satisfied.

关 键 词：中风 中成药适宜人群 LOGISTIC回归分析 

分 类 号：R255.2[医药卫生—中医内科学]