Self-similar properties of long term heart rate variability assessed by discrete wavelet transform

The aim of this study was to test the self-similar properties of 24 h Holter Heart Rate Variability (HRV) data using the Discrete Wavelet Transform (DWT), in a population of hypertensive patients, compared with a normal group. For the DWT analysis we used a 2^nd order Daubechies filter according to theoretical considerations on nonstationarity, border effects and numerical efficiency. We estimated the 1/f spectral exponent as either the slope of the log-log plot of the power spectrum (α-FFT) and the slope of the semilog plot of the DWT details variances vs. the scale (α-DWT). In hypertensive patients, α-FFT was 1.14±0.10 and α-DWT was 1.1±0.09; in normal subjects, α-FFT and α-DWT were 1.08±0.11 and 1.06±0.07, respectively. These results suggest the DWT as a robust and efficient method to characterize long term HRV self-similarity, overcoming the limitations of the classical FFT approach for 1/f processes