基于Logistic回归与决策树模型的老年多重慢病及影响因素分析  

Analyzing the Multiple Chronic Diseases of the Elderly and Its Influencing Factors Based on Logistic Regression and Decision Tree Model

作　　者：董海颖 梁笑笑 何燕 Dong Haiying(College of Public Health,Zhengzhou University,Zhengzhou,P.R.China)

机构地区：[1]郑州大学公共卫生学院社会医学与卫生事业管理学系,河南郑州450001 [2]海南医学院

出　　处：《中国卫生事业管理》2024年第2期208-211,共4页Chinese Health Service Management

基　　金：2021年河南省重点研发与推广专项(科技攻关)项目“基于中国人口死亡模式的动态多水平社会化养老照护研究”(212102310814);2022年郑州大学教授团队助力企业创新驱动发展专项“社区生态宜居老化空间的环境建构”(JSZLQY2022023);2022年郑州大学课程思政教育教学示范课程“社会医学”(2022ZZUKCSZ040)。

摘　　要：目的:应用决策树模型和Logistic回归对老年多重慢病影响因素进行分析,为多重慢病的管理和防控提供依据。方法:采用多阶段分层随机抽样方法于青岛、广州、苏州抽取1273名老年人,进行老年多重慢病的问卷调查,分别建立Logistic回归模型和决策树模型,分析并比较两种分析方法结果的差异性。结果:Logistic回归结果显示年龄、婚姻状况、医疗保险、吸烟是老年多重慢病的影响因素,其中年龄是保护因素(OR<1)。决策树模型显示医疗保险是老年多重慢病的最主要影响因素,其次是吸烟、年龄和婚姻状况。两种模型分析比较结果显示,Logistic回归模型的灵敏度为74.3%,特异度为55.3%;决策树模型的灵敏度为57.2%,特异度为73.3%。结论:研究显示年龄、婚姻状况、医疗保险、吸烟是老年多重慢病的影响因素。结合运用Logistic回归模型和CHAID模型,可以有效筛选老年多重慢病的危险因素,有助于制定针对性措施,加强管理和防控。Objective To analyze the influencing factors of multiple chronic diseases in the elderly by using decision tree mod⁃el and Logistic regression,and provide evidence for the management,prevention and control of multiple chronic diseases.Methods 1273 elderly people were selected from Qingdao,Guangzhou and Suzhou by multi-stage stratified random sampling method to conduct a questionnaire survey on multiple chronic diseases in the elderly.Logistic regression model and decision tree model were established respectively to analyze and compare the differences between the results of these two analysis methods.Results Logistic regression showed that age,marital status,medical insurance and smoking were the influencing factors of multiple chronic diseases in the elderly,in which age was the protective factor(OR<1).The decision tree model showed that medical insurance was the most important factor affecting multiple chronic diseases in old age,followed by smoking,age and marital status.The sensitivity and specificity of Logistic regression model was 74.3%and 55.3%,respectively.The sensitivity and specificity of decision tree model was 57.2%and 73.3%,respectively.Conclusion Age,marital status,medical insurance and smoking are the influencing factors of multiple chronic diseases in the elderly.The combination of Logistic regression model and CHAID model can effectively screen the risk factors of multiple chronic diseases in the elderly,and help formulate targeted measures to strengthen management and prevention and control of multiple chronic diseases in the elderly.

关 键 词：LOGISTIC回归 决策树模型 老年多重慢病 影响因素 

分 类 号：R19[医药卫生—卫生事业管理]