Nonlinear noise reduction for the analysis of heart rate variability signals in normal and heart transplanted subjects

Presents results obtained applying a nonlinear noise reduction method to time series measured from heart rate variability (HRV) signals. Signals generated by nonlinear systems show broad-band spectra which makes it not useful to employ linear filters to obtain noise reduction without introducing signal distortion. The noise reduction method the authors used works directly in the reconstructed system space state. The method is applied to the HRV signals of 9 normal and 6 heart transplanted subjects providing more reliable estimation of D2, Hurst coefficients and α slope of 1/f spectrum. Moreover the noise reduction leads to a better identification of HRV signal complex trajectories that can be clearly represented by delay maps