Development and Validation of a Survival Machine Learning Model for Predicting Major Adverse Cardiovascular Events After Percutaneous Coronary Intervention in Patients With Coronary Artery Aneurysm or Ectasia

Background Patients with coronary artery aneurysm or ectasia (CAAE) remain at risk of major adverse cardiovascular events (MACE) after percutaneous coronary intervention (PCI). We aimed to develop and temporally validate a model for post-discharge MACE. Methods This single-centre retrospective cohort included 574 adults with angiographically confirmed CAAE who underwent PCI and survived to discharge. Patients were chronologically assigned to development (n = 456) and temporal validation (n = 118) cohorts. Missing data were handled using five MICE datasets. LASSO-Cox selected predictors, and six survival algorithms were compared using 10 repeats of stratified five-fold cross-validation. Internal validation used 1,000 patient-level bootstrap resamples with optimism correction; temporal validation was performed without refitting or recalibration. Results During follow-up, 85 development-cohort patients and 12 temporal-validation patients experienced MACE, including 45 and 11 patients, respectively, within 3 years. Seven predictors were retained. Random survival forest (RSF) was selected according to the prespecified rule, with the highest mean out-of-fold C-index (0.761, SD 0.038). In development, the optimism-corrected C-index and 3-year AUC were 0.813 (95% CI 0.782–0.838) and 0.847 (0.806–0.881); corresponding temporal validation estimates were 0.826 (0.734–0.910) and 0.843 (0.742–0.931). Compared with Cox, RSF improved 3-year AUC by 0.098 in development (P = 0.018) and 0.118 in temporal validation (P = 0.032). Conclusions RSF showed good discrimination and identified a high-risk group, with preliminary support from same-centre temporal validation. Calibration and clinical net benefit require confirmation in larger, multicentre cohorts before clinical use.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11572812/v1
Primary Topic
Coronary Interventions and Diagnostics
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Development and Validation of a Survival Machine Learning Model for Predicting Major Adverse Cardiovascular Events After Percutaneous Coronary Intervention in Patients With Coronary Artery Aneurysm or Ectasia

Yuxin Yang, Zhichao Dong, Xiaopeng Dong, Yuhao Liu et al.
5 links
Research Square
Coronary Interventions and Diagnostics
preprint

Development and Validation of a Survival Machine Learning Model for Predicting Major Adverse Cardiovascular Events After Percutaneous Coronary Intervention in Patients With Coronary Artery Aneurysm or Ectasia

Yuxin Yang, Zhichao Dong, Xiaopeng Dong, Yuhao Liu, Bo Zhang
preprint en

Abstract

Background Patients with coronary artery aneurysm or ectasia (CAAE) remain at risk of major adverse cardiovascular events (MACE) after percutaneous coronary intervention (PCI). We aimed to develop and temporally validate a model for post-discharge MACE. Methods This single-centre retrospective cohort included 574 adults with angiographically confirmed CAAE who underwent PCI and survived to discharge. Patients were chronologically assigned to development (n = 456) and temporal validation (n = 118) cohorts. Missing data were handled using five MICE datasets. LASSO-Cox selected predictors, and six survival algorithms were compared using 10 repeats of stratified five-fold cross-validation. Internal validation used 1,000 patient-level bootstrap resamples with optimism correction; temporal validation was performed without refitting or recalibration. Results During follow-up, 85 development-cohort patients and 12 temporal-validation patients experienced MACE, including 45 and 11 patients, respectively, within 3 years. Seven predictors were retained. Random survival forest (RSF) was selected according to the prespecified rule, with the highest mean out-of-fold C-index (0.761, SD 0.038). In development, the optimism-corrected C-index and 3-year AUC were 0.813 (95% CI 0.782–0.838) and 0.847 (0.806–0.881); corresponding temporal validation estimates were 0.826 (0.734–0.910) and 0.843 (0.742–0.931). Compared with Cox, RSF improved 3-year AUC by 0.098 in development (P = 0.018) and 0.118 in temporal validation (P = 0.032). Conclusions RSF showed good discrimination and identified a high-risk group, with preliminary support from same-centre temporal validation. Calibration and clinical net benefit require confirmation in larger, multicentre cohorts before clinical use.

Research Square
First Affiliated Hospital of Dalian Medical University (CN)
Natural Science Foundation of Liaoning Province, National Science and Technology Major Project
Good health and well-being
Coronary Interventions and Diagnostics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon