Does a bidirectional GRU network using Doppler radar improve heartbeat detection accuracy compared to state-of-the-art methods?
A novel bidirectional GRU network using Doppler radar provides highly accurate, contactless heartbeat detection, outperforming existing methods.
Heart rate is one of the most critical and important vital signs in healthcare. While electrocardiography (ECG) is gold-standard procedure for heart rate monitoring, contactless monitoring is preferred in many applications like long-term monitoring. Radar systems enable contactless sensing by measuring small movements on the chest induced by the heart beat. In this paper, we present a machine learning-based method using a bidirectional gated recurrent unit (bi-GRU) network for accurate heartbeat detection. Band-pass filtered in-phase (I) and quadrature (Q) signals in heart sound and pulse wave frequency ranges were fused. The proposed method achieves a high F1 score of 98.06% for heart beat detection, thus outperforming the state-of-the-art method with an F1 score of 95.62% in the resting scenario. In the tilt-up scenario with the tilt table, F1 score is significantly improved by 10%. Besides, a median inter-beat intervals (IBIs) RMSE of only 22.07 ms in the resting scenario is realized.
Lu et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: