首页|基于XGboost学习算法构建模型对缺血性脑卒中合并房颤患者并发肺部感染的预测价值

基于XGboost学习算法构建模型对缺血性脑卒中合并房颤患者并发肺部感染的预测价值

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
目的 探讨基于XGboost学习算法构建模型对缺血性脑卒中合并房颤患者并发肺部感染的预测价值。方法 选择2018年1月-2019年12月南昌大学第二附属医院收治的缺血性脑卒中合并房颤患者1 075例为研究对象，按照7∶3随机分为训练集与验证集，训练集用于构建预测模型，验证集用于评价模型效果;分别采用XGboost和Logistic回归构建缺血性脑卒中合并房颤并发肺部感染的预测模型，通过灵敏度、特异度和受试者特征(ROC)曲线下面积(AUC)评价模型的预测效果。结果 共纳入1 075例缺血性脑卒中合并房颤患者，年龄(74。82±10。85)岁，其中男性554例(51。53％)，肺部感染341例(31。72％);Logistic回归分析模型和XGboost模型的AUC分别为0。671和0。723，灵敏度分别为72。63％和79。01％，特异度分别为68。10％和62。42％，约登指数分别为0。407和0。414。结论 在针对缺血性脑卒中合并房颤患者是否并发肺部感染预测研究中，基于XGboost算法构建的预测模型较Logistic具有良好的预测效能。
Predictive value of a model constructed based on the XGboost learning algorithm for concurrent lung infections in patients with combined atrial fibrillation
OBJECTIVE To investigate the predictive value of a model constructed based on the XGboost learning al-gorithm for concurrent pulmonary infection in patients with ischemic stroke combined with atrial fibrillation(AF).METHODS A total of 1075 patients with ischemic stroke complicated with atrial fibrillation admitted to the Second Affiliated Hospital of Nanchang University from Jan.2018 to Dec.2019 were selected as the study subjects,and they were randomly divided into a training set and a validation set in accordance with 7∶3,with the training set being used to construct the prediction model,and the validation set being used to evaluate the model effect.XG-boost and Logistic regression were used to construct the prediction model for ischemic stroke complicated with at-rial fibrillation and pulmonary infection,respectively.The predictive effect of the model was evaluated by sensitivi-ty,specificity and area under the curve(AUC)of receiver operating characteristic(ROC).RESULTS A total of 1075 patients with ischemic stroke complicated with atrial fibrillation were included,with a mean age of(74.82± 10.85)years old,of whom 554(51.53％)were males and 341(31.72％)had pulmonary infection.The AUC of the Logistic regression analysis model and the XGboost model were 0.671 and 0.723 respectively,with a sensitivity of 72.63％and 79.01％,a specificity of 68.10％and 62.42％,and the Yoden index of 0.407 and 0.414,respective-ly.CONCLUSION In the study of prediction model for ischemic stroke combined with atrial fibrillation,the predic-tion model constructed based on the XGboost algorithm had better predictive efficacy compared with Logistic.

Ischemic strokeAtrial fibrillationPulmonary infectionXGboostLogistic regressionPrediction model

王小曼、韩梦琦、张晓林、俞鹏飞、罗颢文、刘建模、刘松、易应萍

展开 >

南昌大学第二附属医院医疗大数据研究中心,江西南昌 330006

南昌大学公共卫生学院(江西省预防医学重点实验室),江西南昌 330000

南昌大学第二附属医院科技处,江西南昌 330006

缺血性脑卒中 房颤 肺部感染 XGboost Logistic回归 预测模型

国家自然科学基金地区科学基金资助项目科技部国家重点研发计划基金资助项目江西省重点研发计划重点基金资助项目南昌大学第二附属医院院内基金资助项目江西省应用研究培育计划基金资助项目

819606092020YFC200290120223BBH800132021efyB0320212BAG70029

2024

10.11816/cn.ni.2024-230979

中华医院感染学杂志
中华预防医学会 中国人民解放军总医院

中华医院感染学杂志

CSTPCD北大核心
影响因子：1.885
ISSN：1005-4529
年,卷(期)：2024.34(3)
  • 26