Determining heart rate beat-to-beat from smartphone seismocardiograms: Preliminary studies

In the last decade the development of high quality, sensitive and inexpensive accelerometers has been seen, which in combination with low cost computational power provided the reasons for reconsidering analysis of cardiovascular vibrations in clinical practice. Seismocardiography (SCG) is a non-invasive method of analyzing and recording vibrations generated by heart activity and blood motion. Mobile devices offer the possibility to monitor health parameters. Various applications have been proposed for SCG, including HRV (heart rate variability) analysis. Our aim is to determine location of AO points of SCG to achieve heart rate (HR) changes in time. Proposed algorithm consists of calculating total acceleration value, signal preprocessing, peak finding, computing time between consecutive AO peaks and converting to heart rate. Algorithm performance was measured as true positive (TP), false positive (FP), false negative (FN) rates, sensitivity (Se) and positive predictive value (PPV) on 833 beats collected from 4 subjects. We achieved average Se = 0.868 and PPV = 0.737 and in the best case Se = 0.995 and PPV = 0.974. The obtained results are encouraging and indicate the possibility of measuring heart rate beat-to-beat accurately in rest conditions.