Pattern complexity and nonlinear dynamics in RR-sequences

Abstract

The analysis of time series measured from nonlinear signals, may be performed either in the phase space or in the tie-domain. The Largest Lyapunov Exponent (LLE) characterises exponential divergence of trajectories in the phase space; fractal analysis is able to describe the complex pattern of a given time series. To evaluate the relation between the dynamic behavior and pattern complexity of the inherent biological system, RR-interval sequences were derived from 24-hour Holter recordings performed in 55 healthy subjects (37/spl plusmn/4 years, 34 males). Pattern fractal analysis (PFD) was computed on the basis of the measured length and diameter of the signal pattern. and LLE was evaluated by the Wolf algorithm. For each subject, the linear regression between computed PFD and LLE measures over the 24-hour period has been computed, extracting the correlation coefficient and the slope of the PFD vs. LLE relation. The strongest linear correlation between LLE and PFD indicates a light link between the system dynamics and the pattern of the extracted signals. This link suggests the possibility of a direct evaluation of nonlinear dynamics, even over short time intervals, exploiting the computationally less expensive PFD.