Contactless Heart Sound Detection Using Advanced Signal Processing Exploiting Radar Signals

Abstract

Contactless vital signs detection has the potential to advance healthcare by offering precise and convenient patient monitoring. This groundbreaking approach not only streamlines the monitoring process, but also allows continuous, real-time assessment of vital signs, allowing early detection of anomalies and prompt intervention. This paper presents a novel framework for contactless vital sign s detection using continuous-wave (CW) radar and advanced signal processing techniques. We achieved unprecedented precision in capturing 1,261 samples for radar based heart sound waveforms compared to the ground truth ECG signal. Further, our heart sounds method yields highly accurate human heart pulse readings, surpassing previous benchmarks with a mean absolute percentage error (MAPE) of 0.0129 and mean absolute error (MAE) below one (0.8712). In addition, we derive d heart rates from the heart sound waveforms and compare them with conventional radar-derived heart rates and ground truth ECG signal. Through analysis, we identifi ed regions where conventional radar based methods exhibit limitations. Our approach demonstrates minimal errors and superior accuracy across all heart rate states, which can potentially set new standards for noninvasive vital sign monitoring.