Probing temporal correlation in ventricular interbeat intervals during atrial fibrillation with local continuous DFA

Using the method of local Continuous Detrended Fluctuation Analysis (CDFA) we analyze the correlations of ventricular interbeat intervals of patients with Atrial Fibrillation (AF). CDFA yields a local Holder exponent h for a neighborhood around each point in the time series by determining the scaling of fluctuations with window size after detrending. We compare the histograms of Holder exponents for original data with those of randomly shuffled data and find some correlations not only in long-range windows but also at short time scales where interbeat intervals during AF have been believed to be random in nature. Furthermore, we find unique temporal correlation structures to occur only in the heart rate of patients who were in the survivor group when a follow up was conducted at least one year after data acquisition. We conclude that ventricular interbeat intervals during AF contain richer information than previously considered and the study of the local correlations may be useful in predicting mortality of the patients.