Effect of the Average Delay and Mean Connectivity of the Kuramoto Model on the Complexity of the Output Electroencephalograms

Cognitive functions result from the interplay of distributed brain areas operating in large-scale networks. These networks can be modelled with a number of parameters that represent their underlying dynamics. One particularly fruitful model to simulate key aspects of the large-scale brain networks is the Kuramoto model, which simulates the phase evolution of several weakly coupled oscillators that represent the mean oscillatory behavior of different cortical regions. Here, we inspected the dependency of two widespread nonlinear complexity markers, Sample Entropy (SampEn) and Lempel-Ziv Complexity (LZC), on EEG activity generated with a Kuramoto phase model where the time delay and connectivity strength among oscillators varied. We also added different levels of noise to the electroencephalogram (EEG) signals. Our results indicated that both complexity metrics reflected the changes in the delays and global synchrony levels, but we found that SampEn was slightly more sensitive to the state transition and its results were less affected by the presence of noise. These results help in the effort to understand the dynamics of EEG recordings and their relationship to large-scale networks.