The proposed multi-modal wearable heart-risk prediction system integrates physiological signals with machine learning and explainable AI to generate a continuous cardiovascular risk score.
A proposed multi-modal wearable pipeline using machine learning and explainable AI aims to shift cardiovascular monitoring from reactive to personalized, continuous risk prediction.
Heart and circulatory conditions remain a leading cause of death globally, and they often worsen quietly before obvious symptoms appear. Consumer wearables now make it practical to collect physiology around the clock, yet many pipelines still lean on one signal at a time and collapse decisions into coarse “normal vs abnormal” labels that arrive late for prevention. This work outlines a personalized, multi-modal wearable pipeline that jointly uses ECG, photoplethysmography (PPG), heart-rate variability (HRV), blood-oxygen saturation (SpO₂), and sleep-related cues. Machine learning sits on top of adaptive baselines so the model can learn what is typical for a given person, not only population averages, and it emits a continuous risk score from 0 to 100 that is easier to track over days and weeks. Explainable AI (XAI) is folded in so users and clinicians can see which factors moved the score, supporting trust and safer follow-up. Overall, the aim is to shift monitoring from reactive firefighting toward earlier, individualized cardiovascular vigilance.
Fahad et al. (Tue,) conducted a other in Cardiovascular Disease. Personalized Multi-Modal Wearable Data-Based Heart Risk Prediction System was evaluated. The proposed multi-modal wearable heart-risk prediction system integrates physiological signals with machine learning and explainable AI to generate a continuous cardiovascular risk score.