Power Spectral Density of Unevenly Sampled Heart Rate Data *

This work studies the frequency behavior of t h e Lomb method to estimate the power spectral density (PSD) of heart rate (HR) unevenly sampled (UNS) signals. When the UNS can be modeled as uniform plus a random deviation (approximation well satisfied in HR signals), this spectra results in a periodic repa ti t ion of the original continuous t ime spectrum at the mean Nyquist frequency, with a low pass effect affecting the upper bands that depends on the sampling dispersion. In this case the estimation at the base band is unbiased with practically no dispersion. The performance has been tested with real HR signals corroborating the theoretical results. We have found that the Lomb method avoids the classical methods problem of low pass effect due to resampling. We conclude that the Lomb method is more suitable than classical methods for PSD estimation of UNS signals.