Application of Local Scaling Exponent Profile to Heart Rate Variability Analysis

The long-range correlation analysis of time series, such as detrended fluctuation analysis (DFA) and detrending moving-average analysis (DMA), has estimated a scaling exponent with the presence of a sufficiently broad scaling range indicating a linear relation in the log-log plot of the fluctuation function as the necessary condition. Therefore, it was difficult to interpret the results of the scaling analysis when the scaling range did not exist or linearity in the slope estimation broke down. However, a previous study reported that the DFA scaling exponents at very narrow intervals of heart rate variability (HRV) showed clearly different behavior between healthy subjects and patients with congestive heart failure (CHF). In this study, we propose a method to estimate a profile consisting of such local scaling exponents and apply this to long-term HRV analysis. By evaluating the discriminating ability between healthy subjects and CHF patients and between survivor and nonsurvivor groups of CHF patients by receiver operating characteristic (ROC) analysis, we demonstrate the usefulness of the local scaling exponent profile.