Optimal Metrics Investigation for Human Functional States Differences Assessment Based on EEG Signal Analysis

The paper presents an evaluation of the various EEG signal metrics optimality for distinguishing between a person's EEG signal in a state of concentration and in a state of mind-wandering. The metrics are selected based on an analysis of existing research on recognizing and assessing various states similar to concentration. Practical methods for computing the selected metrics of power and entropy in different frequency bands are provided. The study uses a unique dataset containing EEG recordings for functional states of concentration at a point on the center of the forehead and a background state of mind-wandering, including recordings from 17 participants. The metrics optimality is assessed based on the point-biserial correlation coefficient of metric values with functional states, as well as a similar measure based on the difference in interquartile ranges of metric values. The most optimal metrics are power metrics in the alpha, theta, and SMR bands, as well as the signal entropy metric in the range of 0,3 to 30 Hz. These metrics show significant but divergent changes between functional states for almost all participants in the sample.