A modified Hilbert-Huang algorithm to assess spectral parameters in intense exercise

Spectral indices are widely used to assess Heart Rate Variability (HRV) during exercise. HRV signal spectrum comprises two main bands, High Frequency (HF) and Low Frequency (LF), the first related to parasympathetic activity and the second related to both parasympathetic and sympathetic activity. HF and LF powers are mostly obtained by Fast Fourier Transform (FFT) based algorithms, however there is a major problem due to the non-stationary and non-linear properties of the signal. Also, FFT based algorithms usually provide single LF and HF indices for temporal windows of several minutes. In the present study, our aim was to achieve a deeper understanding on the autonomic regulation mechanisms during intense exercise and recovery. For this purpose, we obtained the instantaneous LF and HF indices using a modified version of the Hilbert-Huang (HH) algorithm to track the HRV evolution on eight male amateur triathletes in an All Out Exercise Test (AOET). Both HH-based and FFT-based algorithms revealed severely depressed LF and HF powers during exercise. However, using the FFT the LF/HF ratio was always lower than one during intense exercise, while the mean of the instantaneous LF/HF ratio was lower than one only in one case. The HH-based algorithm allowed a deeper insight about the sympathetic and parasympathetic balance during exercise.