The purpose of this work is to study the effectiveness of the method for analyzing directional couplings between oscillatory systems on test models that reproduce the spectral properties of electroencephalogram (EEG) signals of healthy subjects at rest.

Methods. Eleven EEG recordings from 10 healthy young adults in a supine position were used as experimental signals. Two distinct rhythms were observed in the spectra of most signals:around 0.1 Hz and around 10 Hz. Based on the basic equations of oscillation theory, special test mathematical models were created demonstrating oscillations at frequencies of 0.1 Hz and 10 Hz. A method for analyzing directional couplings, based on empirical modeling of phase dynamics, was tested to identify relationships between 0.1 Hz oscillations in the time series of the proposed test systems. A filtering method and time series length were selected that demonstrated high sensitivity and a low rate of erroneous conclusions.

Results. It is shown that the method makes it possible to correctly identify directional couplings when using bandpass filters 0.02-0.5 Hz and 0.05-0.15 Hz and for time series with a length of at least 70 typical periods. It is shown that an estimate of the mean phase coherence coefficient can be used to diagnose situations of possible erroneous conclusions about the direction of coupling. The application of the obtained results is illustrated by the experimental EEG signals of a healthy subject.

Conclusion.Recommendations are proposed for choosing the filtering method and the length of the time series for which the method of estimating directional couplings works correctly. Using real EEG signals from a healthy volunteer, it was shown that the use of a wider bandpass filter provides additional information about the couplings of low-frequency rhythms of EEG leads. A moving window coupliong analysis showed that the architecture of interaction between some leads can change significantly within 30 minutes.