An Artificial Intelligence based model for predicting long-term all-cause mortality after acute Myocardial Infarction (the AIMI model)

Abstract Background and aims Predicting long-term mortality after acute myocardial infarction (AMI) remains challenging. We aimed to establish an Artificial Intelligence - based model for predicting long-term all-cause mortality after AMI (the AIMI model). Methods AIMI model was employed by RF (Random forest). Individual predictions were visualized by SHAP plots. AIMI model was compared against existing clinical risk scores using time-dependent ROC (receiver operating characteristic) curves, and Kaplan-Meier (K-M) analyses. External validation was also performed at the same way. Brier scores were calculated in validation cohorts. Results We consecutively enrolled 4825 AMI patients underwent emergent coronary angiography or PCI procedures within 24 hours of symptom onset to train and test the AIMI model and 723 AMI patients for external validation. Model incorporated 15 variables achieved robust performance (C-index=0.81). As indicated by AUCs in the test set, AIMI model outperformed GRACE and TIMI risk scores across short-, mid- and long-term periods, especially for long-term prediction (1, 3 and 5 years). K-M curves confirmed precise discrimination between low-, median-, and high-risk groups (all p0.05). External validation confirmed good generalization and robustness for AIMI model (AUCs: 0.88, 0.91, 0.83, 0.75, 0.78 and 0.77 during hospitalization, at 30 days, half-year, 1-year, 2-years and 3-years follow-up; comparisons of K-M curves across three risk groups, all p0.05). Brier scores demonstrated good individual level performance (internal validation cohort: 0.022; external validation: 0.021). Conclusions The AIMI model surpassed traditional methods for long-term all-cause death prediction after AMI. AI-based model demonstrated potential to enhance risk stratification and guide post-discharge management.