Mortality prediction in patients with acute myocardial infarction and prior ischemic stroke: a prospective cohort study with explainable machine learning and SHAP analysis
- 1Department of Cardiology, First Affiliated Hospital of Xinjiang Medical University
- 2School of Public Health, Xinjiang Medical University
- 3Department of Cardiology, People's Hospital of Xinjiang Uyghur Autonomous Region
Abstract
Background: Patients with acute myocardial infarction (AMI) and a prior history of ischemic stroke (IS) represent a high-risk population with limited prognostic tools. This study aimed to develop and validate an interpretable machine learning model for predicting all-cause mortality in AMI patients with prior IS.
Methods: This prospective single-center cohort included 543 consecutive patients with AMI and prior IS who underwent Percutaneous Coronary Intervention (PCI). Candidate predictors were screened using univariable Cox regression, Boruta, and LASSO Cox regression. Six survival models were compared: Cox, elastic-net Cox, random survival forest (RSF), GBM survival, XGBoost survival, and Survival SVM. Internal validation used 100 bootstrap resamples with out-of-bag (OOB) evaluation. Performance was evaluated using Harrell's C-index, IPCW time-dependent AUC, calibration measures, Brier score, and decision-curve analysis. SHAP-based and permutation analyses were used for model interpretation.
Results: During a median follow-up of 3.03 years, 66 patients (12.2%) died. Six predictors were selected: age, hemoglobin, prognostic nutritional index, estimated glomerular filtration rate, atherogenic index of plasma, and monocyte-to-lymphocyte ratio. Although RSF achieved the numerically highest OOB C-index, elastic-net Cox showed comparable discrimination, better calibration, a lower 3-year Brier score, and greater interpretability and was selected as the final model. Its OOB C-index was 0.714 (95% CI 0.616–0.790), and its 1-, 2-, and 3-year IPCW AUCs were 0.745, 0.704, and 0.701, respectively. Age and hemoglobin made the largest contributions to model predictions. Elastic-net Cox showed numerically better performance than GRACE, but statistical superiority in discrimination was not demonstrated.
Conclusions: The six-variable elastic-net Cox model provided moderate discrimination and relatively favorable calibration. Independent external validation and prospective implementation studies are required before clinical use.
Related
- Mortality risk stratification in contemporary myocardial infarction-related cardiogenic shock: development of AMICS-ADMIT prediction model and benchmarking of established scores.
- ECG-Derived Synthetic Left Ventricular Mechanics Predict Mortality After Acute Myocardial Infarction and Fatal Myocardial Infarction in the General Population
- Factors associated with 30-day mortality in patients with post-myocardial infarction ventricular septal rupture: a retrospective cohort study
- Predictive value of combined estimated plasma volume status and neutrophil-to-lymphocyte ratio for cardiovascular mortality in patients with acute myocardial infarction with heart failure
- Comparative Associations of Inflammation and Nutrition-Based Indices with No-Reflow and Mortality After Primary PCI for ST-Elevation Myocardial Infarction
BibTeX
@article{han2026mortality,
title = {{Mortality prediction in patients with acute myocardial infarction and prior ischemic stroke: a prospective cohort study with explainable machine learning and SHAP analysis}},
author = {Congcong Han and Abudureheman Muladili and Ning Song and Chunfang Shan and Fen Liu and Xiaomei Li and Yining Yang and Qian Zhao},
journal = {Frontiers in Cardiovascular Medicine},
year = {2026},
month = oct,
doi = {10.3389/fcvm.2026.1949226},
url = {https://www.frontiersin.org/articles/10.3389/fcvm.2026.1949226/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Mortality prediction in patients with acute myocardial infarction and prior ischemic stroke: a prospective cohort study with explainable machine learning and SHAP analysis
AU - Congcong Han
AU - Abudureheman Muladili
AU - Ning Song
AU - Chunfang Shan
AU - Fen Liu
AU - Xiaomei Li
AU - Yining Yang
AU - Qian Zhao
PY - 2026
DA - 2026/10/27
JO - Frontiers in Cardiovascular Medicine
DO - 10.3389/fcvm.2026.1949226
UR - https://www.frontiersin.org/articles/10.3389/fcvm.2026.1949226/full
AB - Background: Patients with acute myocardial infarction (AMI) and a prior history of ischemic stroke (IS) represent a high-risk population with limited prognostic tools. This study aimed to develop and validate an interpretable machine learning model for predicting all-cause mortality in AMI patients with prior IS. Methods: This prospective single-center cohort included 543 consecutive patients with AMI and prior IS who underwent Percutaneous Coronary Intervention (PCI). Candidate predictors were screened using univariable Cox regression, Boruta, and LASSO Cox regression. Six survival models were compared: Cox, elastic-net Cox, random survival forest (RSF), GBM survival, XGBoost survival, and Survival SVM. Internal validation used 100 bootstrap resamples with out-of-bag (OOB) evaluation. Performance was evaluated using Harrell's C-index, IPCW time-dependent AUC, calibration measures, Brier score, and decision-curve analysis. SHAP-based and permutation analyses were used for model interpretation. Results: During a median follow-up of 3.03 years, 66 patients (12.2%) died. Six predictors were selected: age, hemoglobin, prognostic nutritional index, estimated glomerular filtration rate, atherogenic index of plasma, and monocyte-to-lymphocyte ratio. Although RSF achieved the numerically highest OOB C-index, elastic-net Cox showed comparable discrimination, better calibration, a lower 3-year Brier score, and greater interpretability and was selected as the final model. Its OOB C-index was 0.714 (95% CI 0.616–0.790), and its 1-, 2-, and 3-year IPCW AUCs were 0.745, 0.704, and 0.701, respectively. Age and hemoglobin made the largest contributions to model predictions. Elastic-net Cox showed numerically better performance than GRACE, but statistical superiority in discrimination was not demonstrated. Conclusions: The six-variable elastic-net Cox model provided moderate discrimination and relatively favorable calibration. Independent external validation and prospective implementation studies are required before clinical use.
SN - 2297-055X
ER -