Spectral analysis of heart rate variability

Abstract

Abstract Spectral analysis (SA) has been extensively applied to the assessment of heart rate variability. Traditional methods require the transformation of the original non-uniformly spaced electrocardiogram RR interval series into regularly spaced ones using interpolation or other approaches. The Lomb-Scargle (L-S) method uses the raw original RR series, avoiding different artifacts introduced by traditional SA methods, but it has been scarcely used in clinical settings. An RR series was recorded from 120 healthy participants (17–25 years) of both genders during a resting condition using four SA methods, including the Classic modified periodogram, the Welch procedure, the autoregressive Burg method and the L-S method. The efficient implementation of the L-S algorithm with the added possibility of the application of a set of options for the RR series pre-processing developed by Eleuteri et al., and also the results obtained in this study, show that the L-S method can be a good choice for future clinical studies. The L-S method seems particularly useful when the heart rates of studied participants will be below 60 or over 120 beats per minute. Nevertheless, it is important to the development of a smoothing procedure for the L-S spectra to avoid the picky behavior of the L-S power spectrum. The implementation of the L-S algorithm used in this study has been recently published by other authors included in our references, and brings some particular filtering features. The results obtained, comparing the four spectral methods, show that this implementation seems particularly useful when the heart rates of studied participants will be below 60 or over 120 beats per minute. Nevertheless, it is important to recommend for all existing L-S software implementations, the development of a smoothing procedure to avoid the picky behavior of the L-S power spectrum.