Heartbeat Detection Using Three-Axial Seismocardiogram Acquired by Mobile Phone

Abstract

Recent studies have demonstrated the possibility to acquire the heartbeat induced vibrations resembling the seismocardiogram (SCG) by the mobile phone accelerometer (m-SCG). As most of ECG-free beat detection algorithms are based on processing one m-SCG component only (e.g. head-to-foot), our aim was to propose an alternative and robust method to extract beat-to-beat systolic complex (SC) from three-axial m-SCG and test its occurrence compared to R peak of ECG. Eleven male subjects (mean age 22±3) were recruited and m-SCG signal (50 Hz) was acquired for 5 minutes by a smartphone positioned on the chest. Newtonian equations of kinematics and continuous wavelet transform were applied to the m-SCG signal to detect beat-by-beat SC occurrence. Data from one subject were discarded due to low quality signal (feasibility 91%). High values of sensitivity (0.995), positive predictive value (0.974) and accuracy (0.96) were obtained. The explored heart rate range (45-101 bpm) was quite wide and acceptable limits of agreement (±1 - ±6 bpm) were achieved.