Arrhythmia detection from heart rate variability by SDFA method

Abstract

The smoothed detrended fluctuation analysis (SDFA) based on DFA and the principal of wavelet shrinkage procedures is a scaling analysis method to represent the correlation properties of a time series. Since there is not a specific rule for the choice of the numbers of regressors in SDFA Method, we present here an asymptotic optimal choice. We carried out some Monte Carlo simulations on fractional Gaussian noise (FGN) models, to investigate the effect of the numbers of regressors in SDFA Method. We analyze the long dependence property in view of the SDFA method to compare 10 healthy and 10 unhealthy (with cardiac arrhythimia) RR time series randomly selected from databases of the PhysioBank. It is proposed that utilizing Hurst estimator by SDFA method, as an additional diagnostic tool may provide an indication of cardiac arrhythmia.

Keywords: Arrhythmia detection; Hurst estimator; Smoothed detrended fluctuation analysis.