A statistical comparison of signal representations and spectral estimators of heart rate variability in hypertensive patients

Abstract

The paper compares different methods for the estimation of the spectrum in the heart rate variability time series. Studying 32 hypertensive patients it is shown that the low-pass filtering of event series technique drives to an over estimation of the higt frequency component of the heart beat when compared with the results of both cubic spline interpolation and interval tachogram techniques. No differences are founded between the spectrum estimation obtained by autoregressive models and non parametric methods like the Fourier transform.