CLASSIFICATION OF EYE STATE BASED ON EEG DATA USING RECURRENCE ANALYSIS

The relevance of this study is driven by the growing interest in portable EEG devices and the need to develop efficient algorithms for analyzing brain activity with limited technical resources. This paper addresses the problem of classifying brain states based on elec-troencephalography (EEG) data to distinguish between two specific states: relaxation and concentration. The classification of open and closed eyes is examined, as eye closure is asso-ciated with increased relaxation. A classification method based on the quantitative analysis of recurrence plots, which is one of the approaches of chaos theory, is proposed and compared with traditional brain rhythm analysis. Experimental results showed that the recurrence anal-ysis method outperforms spectral analysis in classification accuracy, particularly for the O1 point, where accuracy increased from 86% to 95%. The optimal parameters for phase space reconstruction were determined: delay 25 ms and dimension of the embedding space 4, which are consistent with the spectral characteristics of the signal. Feature importance analysis re-vealed that the most significant parameters for classification are entropy, the length of white vertical and diagonal lines in recurrence plots, as well as determinism and laminarity. The obtained results may be useful for developing EEG analysis algorithms in portable devices and applications in the fields of brain-computer interfaces and cognitive training.