Machine learning prediction of major adverse cardiovascular events following endovascular aneurysm repair in the elderly with abdominal aortic aneurysm
10.3760/cma.j.issn.0254-9026.2025.12.009
- VernacularTitle:老年腹主动脉瘤腔内修复术后主要心血管不良事件机器学习预测模型
- Author:
Yaming ZHOU
1
;
Ning ZHAO
;
Wenxin ZHAO
;
Yixuan WANG
;
Zhiyuan WU
;
Dajie SUOLANG
;
Zuoguan CHEN
;
Yongpeng DIAO
;
Ciren PUBU
;
Yongjun LI
Author Information
1. 西藏自治区人民医院血管外科，西藏 850000
- Publication Type:Journal Article
- Keywords:
Aortic aneurysm, abdominal;
Models, statistical;
Endovascular procedures;
Adverse effects;
Machine learning
- From:
Chinese Journal of Geriatrics
2025;44(12):1674-1681
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish the predictive model for major adverse cardiovascular events(MACE) following endovascular repair in elderly patients with abdominal aortic aneurysm(AAA).Methods:The clinical data and postoperative MACE were retrospectively collected from elderly patients with AAA who underwent their first endovascular aneurysm repair(EVAR)in Beijing Hospital and Tibet Autonomous Region People's Hospital between January 2016 and December 2023.Patients were randomly divided into training and validation cohorts at a ratio of 7∶3.Predictive models were using logistic regression, LASSO regression, random forest, linear discriminant analysis, na?ve Bayes, k-nearest neighbor algorithm, support vector machine, decision tree, and AdaBoost.Models were evaluated using receiver operating characteristic(ROC)curves.Results:A total of 171 elderly AAA patients were enrolled, aged 60 to 94 years(mean 73.0 ± 7.5 years), of whom 145 were male.MACE occurred after EVAR in 30 patients(17.5%). LASSO regression identified monocyte count, history of coronary artery disease, the ratio of maximum AAA diameter to body mass index(DBR), neutrophil-lymphocyte count ratio(NLR), and age as significant predictors, yielding an area under the ROC curve(AUC)of 0.816.Logistic regression achieved an AUC of 0.813 in the training cohort and 0.772 in the validation cohort.Among all models, AdaBoost demonstrated the best performance, with an AUC of 0.92 in the validation cohort.Conclusions:Age, monocyte count, DBR, NLR and creatinine could predict the occurrence of MACE after EVAR in AAA patients.The AdaBoost model provides the most accurate prediction of postoperative MACE.
