Auscultawear: A Wearable Audio-Based Monitoring Platform for Heart and Respiratory Rates

Most commercial wearable health devices are unable to measure both heart and respiratory rate simultaneously as they rely on photoplethysmography (PPG). Audio-based approach is a promising development in wearable devices as they can detect and measure both heart and respiratory rate simultaneously. In this study, a platform using an nRF52840 MCU and MAX4466 electret microphone is used to record physiological signals. These signals are sent to a mobile device where a bandpass filter isolates the target frequency of physiological signals. The signal undergoes wavelet thresholding and decomposition using the db6 wavelet, which shows a similar structure to the S1 and S2 signals in heart sounds, to further isolate and denoise the targeted frequencies. A Hilbert transform is applied to get the envelope of the system, which will then pass through a customized peak detection algorithm. The overall device was tested against a pulse oximeter (PPG), a smart watch (EKG), and a KardiaMobile<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">©</sup> EKG Monitor. Results show the device achieved the best heart rate accuracy at FS1 (upper right intercostal space auscultation point) and FS4 (bottom left intercostal space, apex of the heart), with MAE scores of 5.23±1.79 BPM and 5.18±6.58 BPM, respectively. Further validation at FS1 yielded an MAE of 4.58±2.54 BPM on an individual with normal BMI, reinforcing FS1 as the optimal auscultation point for HR monitoring. The most accurate RR measurement was at BS1 with an MAE of 1.98±1.21 BPM. Power profiling estimates a battery life of 68.979 hours. Auscultawear demonstrates potential as a reliable, low-cost wearable for real-time monitoring of HR and RR. However, these findings are preliminary, and additional refinements and large-scale validation across diverse demographic groups are necessary to establish the system’s robustness and generalizability.