Multiscale entropy and fluctuation analyses of complex signals

A coarse-graining procedure that averages the original dataset over non-overlapping time windows of increased duration was earlier proposed as part of the multiscale entropy (MSE) computation algorithm. This idea of coarse-grained time series can be used to quantify other features of time series. Here we discuss the application of two methods, MSE and an extended version of detrended fluctuation analysis (DFA) to coarse-grained datasets associated with both simulated and experimental signals. We show that the coarse-graining procedure can improve the diagnostics of changes in the system dynamics performed using extended DFA compared to the MSE method.