The direct power spectral estimation of unevenly sampled cardiac event series

Abstract

The application of the Lomb-Scargle periodogram to heart rate variability (HRV), which makes spectral analysis on unevenly sampled data allowable, is considered. To this end, the Lomb-Scargle periodogram was applied to various cardiac event series from IPFM simulation and real ECG signals. To assess the performance of the Lomb-Scargle periodogram, the resultant power spectrum was compared with that obtained by classical periodogram analysis on the evenly-spaced time series derived from a cardiac event series. In conclusion, it has been shown that the Lomb-Scargle periodogram is very effective in the power spectral analysis of HRV, especially in the presence of arrhythmia and/or dropouts of cardiac events.