Abstract

The rising prevalence of cardiometabolic disorders necessitates proactive and real-time monitoring solutions to enable early detection, risk stratification, and timely intervention. Wearable biosensors, when integrated with edge artificial intelligence (Edge AI), offer a transformative approach for continuous, unobtrusive health monitoring by enabling real-time signal processing, energy-efficient analytics, and privacy-preserving decision-making directly on the device. This review systematically explores the current landscape of wearable biosensor systems tailored for cardiometabolic risk monitoring, focusing on three critical components: hardware architecture, signal preprocessing techniques, and predictive modeling pipelines. We examine biosensing modalities such as ECG, PPG, and electrochemical analytes, and assess their integration into low-power hardware platforms for edge deployment. The paper further reviews denoising, normalization, and feature extraction techniques optimized for on-device signal quality enhancement, followed by an evaluation of machine learning and deep learning models suited for edge inference. Challenges such as power constraints, model compression, latency, and data variability are discussed alongside recent advancements in federated learning, neuromorphic computing, and personalized risk modeling. The paper concludes by identifying future research directions toward scalable, secure, and interpretable edge-AI-enabled biosensor platforms for precision cardiometabolic health management.