Hearable Heartbeats: Recording the Heart Rate from the Ear

A single physiological measurement reflects the momentary physiology under the given circumstances. Longitudinal measurements of a key physiological health indicator, such as heart rate, provide a more comprehensive picture of the physiological condition compared to a single measurement. Recently, a cardiovascular phenomenon synchronized to the heartbeat was recorded from the ear using body-coupled microphones. This indicates that heart rate may be estimated using body-coupled microphones. The present study explores the feasibility of extracting time-varying heart rates using bodycoupled microphones in the ears. METHOD: 10 subjects were recorded in a supine position for 5 minutes, with a bodycoupled microphone (BCM) mounted in each ear. Heart rate (HR) was estimated from the BCM recordings using the PYIN pitch detection algorithm. Since PYIN is traditionally applied to speech data, a grid search was conducted to optimize hyperparameters for BCM-based HR estimation. The HR estimates were systematically compared to electrocardiography-based HR estimations in terms of detection rate, error distribution, and root mean square error (RMSE). RESULTS: In 9 out of 19 recorded ears, the RMSE was below 4 beats per minute (BPM). However, 4 out of 10 subjects exhibited both low detection rates and high RMSE, indicating substantial variability in performance across individuals. DISCUSSION: The study revealed considerable inter-ear variability in both the detection rate and the HR error distribution. This variability likely reflects differences in how well cardiovascular activity is represented in the BCM signal, as well as the algorithm's ability to extract the periodicity of the signal reliably. The error distributions suggest that BCM-based HR estimation is unbiased. CONCLUSION: These findings demonstrate the feasibility of HR estimation using BCM in the ear.Clinical Relevance-Ear-level sensing is an emerging technology with broad applications in health monitoring. This study demonstrates the feasibility of enhancing ear-centered sensing devices with BCM-based cardiovascular monitoring capabilities.