Effects of a Spectral Window on Frequency Domain HRV Parameters

Heart rate variability (HRV) can provide physiological information about the autonomic nervous system. For the HRV spectral analysis, spectral windows are applied as weighting functions to solve the spectral leakage problem. The present study aimed to investigate HRV spectra with respect to spectral windows and to determine the frequency characteristics of an optimal window. Two short-term recordings, comprising twelve 5-minute and twelve 2-minute segments created from the one-hour HRV dataset, were analyzed. The HRV indices with respect to three different windows (Hanning, Hamming, and Blackman) were compared with reference to the Hanning window used widely. Hanning-Blackman window showed significant differences in VLF (t=-18.341 with p < 0.0001) for the 5-minute dataset. However, the Hanning-Hamming window showed no significant difference in all frequency bands for both 5-minute and 2-minute datasets. We suggest a low sidelobe related to amplitude accuracy and a narrow width of the main lobe as the frequency characteristics.