AI-Augmented Remote Patient Monitoring in Cardiovascular Care: A Multi-Model Performance Assessment

Abstract

Cardiovascular diseases (CVDs) remain the leading cause of global mortality, necessitating timely diagnosis, continuous monitoring, and rapid clinical intervention. Remote Patient Monitoring (RPM) powered by Artificial Intelligence (AI) is emerging as a transformative model for managing cardiac patients outside clinical settings. This study evaluates the accuracy, predictive capacity, interpretability, and clinical reliability of multiple AI models integrated into RPM systems for cardiovascular care. A multi-model assessment was conducted using random forest, gradient boosting, long short-term memory (LSTM) networks, and transformer-based architectures on ECG, heart-rate variability, oxygen saturation, and blood-pressure datasets. The study employed 5 evaluation dimensions—signal quality interpretation, anomaly detection sensitivity, short-term risk prediction accuracy, latency, and clinical usefulness. Results indicate that LSTM and transformer models outperform traditional algorithms in beat-to-beat risk prediction, while ensemble methods show superior generalizability in diverse patient conditions. The research concludes with a policy framework and implementation roadmap for AI-driven RPM adoption.