Feasibility study for beat-to-beat heart rate detection by smartphone's accelerometers

Our aim was to test the feasibility of measuring beat-to-beat heart rate from vibrations of the body due to the cardiac activity by smartphone accelerometers. The signal recorded by the accelerometers has a similar morphology respect to the ballistocardiographic signal (BCG). This preliminary study was conducted on one healthy volunteer, with the ECG signal simultaneously acquired as reference. Systolic waves (H, I and J) on the longitudinal (head-foot) acceleration component were automatically detected and used to compute the beat-to-beat duration series (HH, II and JJ). Correlation and Bland-Altman analysis were used to test their accuracy against conventional RR series extracted from the ECG. All series showed high correlation and accuracy with RR, with best results obtained for JJ series (r
²
>0.90 CI = ±25 ms), despite the limited time resolution of the accelerometric signal. Smartphone's SCG-derived beat-to-beat quantification of heart rate appears feasible and accurate, thus opening new possibilities in using this signal for self-monitoring.