摘要
目的基于随机森林模型和LASSO回归探究高血压住院患者医疗费用及其影响因素。方法选取2018年10月至2022年4月广西医科大学第二附属医院出院的1087例高血压患者为研究对象,从医院信息系统收集患者基本特征和住院信息等资料。采用随机森林模型和LASSO回归筛选具有统计学意义的变量纳入多重线性回归,分析医疗费用的影响因素,并进行敏感性分析。结果高血压住院患者医疗费用为6916(5069,10183)元。已婚患者的医疗费用高于其他婚姻状况患者[7075(5247,10276)元比5963(4406,9214)元](P<0.01),其他住院科室患者医疗费用高于心血管内科患者[7565(5345,12478)元比6728(4837,9280)元](P<0.01),转科室患者医疗费用高于未转科室患者[13854(8654,30625)元比6840(5004,9905)元](P<0.01),有主诊断伴随并发症患者医疗费用高于无主诊断伴随并发症患者[12545(8408,29940)元比6823(4989,9494)元](P<0.01),不同年龄、医保类型、住院时间患者的医疗费用比较差异有统计学意义(P<0.01)。当LASSO回归lambda值为1se时,筛选出4个影响因素,经随机森林对自变量重要性评分排序,前4位自变量分别为住院时间、年龄、是否有主诊断伴随并发症、是否转科室。多重线性回归分析显示,住院时间、年龄、是否有主诊断伴随并发症、是否转科室是高血压患者医疗费用的主要影响因素(P<0.01)。敏感性分析结果显示,住院时间增加、年龄增大、有主诊断伴随并发症、转科室治疗与较高医疗费用有关(P<0.05),与原多重线性回归模型结果趋势一致。结论控制高血压患者人群的疾病经济负担时,可考虑从缩短住院时间、增强身体素质以降低患病风险、延缓并发症发生等方面采取相应的措施。
Objective To explore the medical expenses and the influencing factors in inpatients with hypertension based on random forest model and LASSO regression.Methods A total of 1087 inpatients with hypertension were included from the Second Affiliated Hospital of Guangxi Medical University from Oct.2018 to Apr.2022,their data including basic characteristics and hospitalization information were collected from the Hospital Information System.Random forest model and LASSO regression were used to screen statistically significant variables into multiple linear regression to analyze the influencing factors of medical expenses,and sensitivity analyses were carried out.Results The medical expenses in the hypertension inpatients was 6916(5069,10183)yuan.The medical expenses of the married patients were higher than that of the patients with other marital status[7075(5247,10276)yuan vs 5963(4406,9214)yuan](P<0.01),and the medical expenses of the patients in other inpatient departments were higher than that of the cardiovascular patients[7565(5345,12478)yuan vs 6728(4837,9280)yuan](P<0.01),the medical expenses of the patients who changed departments were higher than those who did not[13854(8654,30625)yuan vs 6840(5004,9905)yuan](P<0.01).The medical expenses of the patients with comorbidities associated with the primary diagnosis were higher than that of the patients without comorbidities associated with the primary diagnosis[12545(8408,29940)yuan vs 6823(4989,9494)yuan](P<0.01),and there were statistically significant differences in the medical expenses of patients with different ages,types of medical insurance and length of hospitalstay(P<0.01).4 variables were selected from LASSO regression when the lambda value was 1se.According to the importance score ranking of the independent variables by random forest analysis,,the top 4 variables were length of hospital stay,age,presence of comorbidities associated with the primary diagnosis,and transfer to another department.Multiple linear regression analysis showed that the number of le
作者
谭艳粒
张峥嵘
谢红宁
韦咏
邓旭
陆雅
庞翠军
TAN Yanli;ZHANG Zhengrong;XIE Hongning;WEI Yong;DENG Xu;LU Ya;PANG Cuijun(Disease Management Center,the Second Affiliated Hospital of Guangxi Medical University,Nanning 530007,China)
出处
《医学综述》
CAS
2024年第19期2321-2326,2331,共7页
Medical Recapitulate
基金
广西医疗卫生适宜技术开发与推广应用项目(S2018041)
广西壮族自治区卫生和计划生育委员会自筹经费科研课题(Z20170079)
广西壮族自治区卫生健康委自筹经费科研课题(Z-A20230682,Z20191064)。
关键词
高血压
医疗费用
影响因素
随机森林
LASSO回归
Hypertension
Medical expenses
Influencing factors
Random forest
LASSO regression
