Hypertension Risk Screening Using Long-Term Photoplethysmogram and Ballistocardiograph Measurements from a Smartwatch

The ballistocardiogram (BCG), which captures mechanical vibrations caused by cardiac activity, provides complementary insights into cardiovascular function. This study investigated the potential of combining wearable BCG and photoplethysmogram (PPG) signals for cuffless hypertension screening in longitudinal, real-world settings. We proposed a novel multi-modal framework leveraging BCG and PPG to systematically evaluate two feature categories: (1) temporal markers, such as pulse transit time (PTT), derived from intersignal time delays, and (2) waveform morphological characteristics unique to each modality. A model-based feature selection approach identified blood pressure (BP)-sensitive parameters, which were fused to optimize hypertension risk prediction. The framework was trained and validated on a large-scale longitudinal dataset comprising 17,032 signal segments from 177 subjects. Experimental results demonstrated that integrating BCG and PPG signals significantly enhanced hypertension detection accuracy. Models incorporating PTT features outperformed those without, while the multi-modal approach achieved superior accuracy compared to single-modality models.Clinical relevance- This confirms the feasibility of smart watches for cuffless, non-invasive, and continuous blood pressure monitoring. Smart watches are expected to become an auxiliary tool for hypertension screening and cardiovascular risk assessment in home environments.