Detrended Fluctuation Analysis (DFA) and R-R Interval variability: A new linear segmentation algorithm

The detrended fluctuation analysis (DFA) [1] method is used to quantify the fractal-like scaling properties of the variability of cardiac parameters, i.e. R-R interval data. DFA has proved to be a useful index in predicting survival in heart failure. Several authors have proposed to break the numerical series in two zones with linear slopes.. The breakpoint between segments is empirically situated at log n equals 1,1. We have used the DFA method to process records of passive head up tilt (H.U.T) test done to patients who have suffered one or more faint episodes. Slopes of numerical series obtained from real signals neither change at a specific point, nor have only one breakpoint, especially if they correspond to pathological records. On the contrary some of them present abrupt changes in slope. This fact could be hidden in the traditional computation if changes in slope have different sign, but are detected in our approach. A method that tracks the DFA function, detect breakpoints, and obtain a continuous set of lines between them, and their corresponding slopes, is proposed.