ARTIFICIAL INTELLIGENCE–BASED RISK STRATIFICATION MODELS FOR PREDICTING ACUTE CARDIAC EVENTS

Abstract

Cardiovascular diseases remain a leading cause of global morbidity and mortality, necessitating more precise and individualized risk prediction strategies. This study presents a comprehensive experimental evaluation of artificial intelligence–based cardiovascular risk stratification models using a mixed-methods framework that integrates quantitative performance assessment with qualitative interpretability analysis. Multimodal datasets encompassing electronic health records, physiological signals, imaging-derived features, and high-dimensional clinical variables were analyzed using classical machine learning, deep learning, and hybrid AI architectures. Comparative results demonstrated that advanced fusion, transformer-based, and quantum-enhanced models achieved significantly higher discriminative performance, improved calibration, reduced probabilistic error, and enhanced robustness under stochastic perturbations compared with traditional approaches. Entropy-based analyses revealed superior uncertainty management, while three-dimensional latent risk visualizations captured complex nonlinear disease dynamics. Hybrid convergence and sensitivity analyses further confirmed stable optimization and adaptive learning behavior across iterative training cycles. Collectively, the results indicate that AI-driven models provide more accurate, reliable, and clinically actionable cardiovascular risk predictions than conventional methods. These findings support the translational potential of artificial intelligence as an advanced decision-support framework for early disease detection, personalized cardiovascular care, and optimized clinical risk management.