A Temporal–Spatial Feature Fusion Network for Accurate Non-Contact Blood Pressure Measurement via Radar

Non-contact blood pressure (BP) monitoring offers a comfortable and uninterrupted means of BP assessment, free from the constraints of physical contact. A core challenge in radar-based BP monitoring is the extraction of weak BP-related information from radar signals, which significantly affects both the accuracy and real-time performance of BP prediction models. To address this challenge, we focus on waveform features and temporal continuity, proposing a Temporal-Spatial Feature Fusion Network (TSFN) framework for radar-based BP prediction. The TSFN architecture integrates three components: Residual Networks (ResNet) for the extraction of detailed waveform features, Gated Recurrent Units (GRU) for capturing continuous temporal dependencies, and Multiple Head Attention (MHA) to focus on critical information. To enhance the model's robustness, a Pseudo-Huber loss function was employed to refine the optimization process, providing a smoother gradient transition and improved stability. Evaluations demonstrated impressive accuracies, with mean errors of 0.24±6.78 mmHg for systolic blood pressure (SBP) and 0.25±5.13 mmHg for diastolic blood pressure (DBP). These outcomes meet the standards set by the British Hypertension Society (BHS) for grade ‘A’ benchmarks for SBP and DBP measurements. Notably, the TSFN model avoids the need for complex feature engineering, demonstrating its effectiveness in monitoring BP fluctuations across diverse physiological states at 2s intervals. This feature highlights its potential applicability in real-time monitoring systems. Furthermore, using our proposed TSFN framework, we have validated various combinations of temporal and spatial feature extraction networks. Our findings promise a significant advancement for continuous, non-contact BP monitoring with radar technology.