Template-Based Extraction of Interbeat Intervals from in-Ear Heartbeat Sounds

Recent advancements in hearable technology have led to the development of multifunctional in-ear devices equipped with microphones that can improve audio experiences, assist with hearing and enhance speech communication. This paper introduces a template-based algorithm for extracting interbeat intervals (IBIs) from heartbeat sounds captured by acoustic in-ear microphones. The proposed method involves generating a template using the median waveform of the heartbeat sound, using this template for correlation-based peak detection, and detecting anomalies in the resulting peaks. The algorithm is evaluated on 39 participants in silence and noise, with ground truth provided by concurrent electrocardiography measurements. Results demonstrate that the template-based approach improves the accuracy of IBI extraction, achieving 4.68 milliseconds mean absolute error using a 300-second analysis window, which is an improvement of 55.13% from the benchmark method found in the literature. This study highlights the potential of template-based algorithms for heart rate variability analysis using acoustic in-ear microphones, broadening the applications of hearable technology for continuous non-invasive health monitoring.