Statistical measures derived from the correlation integrals of physiological time series
D Kh Ivanov, Harald A. Posch, Christian Stumpf
Abstract
D Kh Ivanov, Harald A. Posch, Christian Stumpf
Abstract
In this work correlation integrals are used for the analysis of various EEG signals from rabbits in resting states and under the influence of an anesthetic. The comparison with surrogate data reveals nonlinear dynamics in all of the time series. Our attempt to determine the correlation dimension D(2) by the modified algorithm of Theiler [Phys. Rev. A 34, 2427 (1986)] failed since no saturation is reached with increasing embedding dimension. The hypothesis of low-dimensional chaos turns out to be inconsistent with our results, but we can still distinguish, at least qualitatively, between different states of brain dynamics. A quantitative characterization of the time series is possible by defining correlation parameters P(a) derived from correlation integrals reflecting also autocorrelation of the signal. (c) 1996 American Institute of Physics.
OpenAlex reports 21 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In this work correlation integrals are used for the analysis of various EEG signals from rabbits in resting states and under the influence of an anesthetic. The comparison with surrogate data reveals nonlinear dynamics in all of the time series. Our attempt to determine the correlation dimension D(2) by the modified algorithm of Theiler [Phys. Rev. A 34, 2427 (1986)] failed since no saturation is reached with increasing embedding dimension. The hypothesis of low-dimensional chaos turns out to be inconsistent with our results, but we can still distinguish, at least qualitatively, between different states of brain dynamics. A quantitative characterization of the time series is possible by defining correlation parameters P(a) derived from correlation integrals reflecting also autocorrelation of the signal. (c) 1996 American Institute of Physics.
Key concepts: Correlation dimension, Autocorrelation, Correlation integral, Correlation, Statistical physics, Series (stratigraphy), Surrogate data, Dimension (graph theory)