Changes in the Hurst Exponent of Heart Rate Variability during Physical Activity

Abstract

We examine fractal scaling properties of heart rate variability using detrended fluctuation analysis (DFA), during physical activity in healthy subjects. We analyze 11 records of healthy subjects, which include both usual daily activity and experimental exercise. The subjects were asked to ride on a bicycle ergometer for 2.5 hours, and maintained a heartbeat interval of 500–600 ms. In order to estimate the long‐range correlation in the series of heartbeat intervals during controlled physical activity, we apply DFA to the data set with the third‐order polynomial trend removed. For all records during exercise, we observe a characteristic crossover phenomenon at ≈ 300 beats. The scaling exponent in the range > 300 beats (> 3 minutes) during exercise decreases and tends to be closer to white noise (≈ 0.5), which corresponds to uncorrelated behavior. The long‐range scaling exponent during exercise is significantly lower than that during daily activity in this range. Contrary to the currently held view, our results indicate a breakdown in long‐range correlations and 1/f‐like scaling, rather than the increase in the Hurst exponent characteristic of a (congestive) increase in afterload and observed, e.g., in congestive heart failure (CHF) patients. Further, our results suggest an increased load imbalance induced departure from critical‐like behavior, which has recently been reported in healthy human heart rate during daily activity.