Analysis of the human electroencephalogram with methods from nonlinear dynamics
We apply several different methods from nonlinear dynamical systems to the analysis of the degree of temporal disorder in data from human EEG. Among these are methods of geometrical reconstruction, dimensional complexity, mutual information content, and two different approaches for estimating Lyapunov characteristic exponents. We show how the naive interpretation of numerical results can lead to a considerable underestimation of the dimensional complexity. This is true even when the errors from least squares fits are small. We present more realistic error estimates and show that they seem to contain additional, important information. By applying independent methods of analysis to the same data sets for a given lead, we find that the degree of temporal disorder is minimal in a ''resting awake'' state and increases in sleep as well as in fluroxene induced general anesthesia. At the same time the statistical errors appear to decrease, which can be interpretated as a transition to a more uniform dynamical state. 29 refs., 10 figs.
- Research Organization:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States); Observatoire de Nice, 06 (France); Goettingen Univ. (Germany, F.R.)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 7188815
- Report Number(s):
- LA-UR-86-3744; CONF-8609205-1; ON: DE87002928
- Resource Relation:
- Journal Volume: 36; Conference: Conference on temporal disorder in human oscillatory systems, Bremen, F.R. Germany, 8 Sep 1986; Other Information: Portions of this document are illegible in microfiche products
- Country of Publication:
- United States
- Language:
- English
Similar Records
No short-term effects of digital mobile radio telephone on the awake human electroencephalogram
Cerebral blood flow during paroxysmal EEG activation induced by sleep in patients with complex partial seizures