Intelligent Method for Human Concentration Recognition Based on EEG Sensor Data

The study examines the applicability of various EEG signal metrics for functional states of concentration and mind-wandering recognition. The metrics were selected based on related work analysis in the topic of human functional states (similar to concentration) recognition. We describe algorithms for calculating the chosen power and entropy metrics across different frequency bands. The study utilizes a dataset containing EEG recordings of functional states, such as concentration on a point in the center of the forehead and a baseline mind-wandering state, including data from 17 participants. The applicability of the metrics was assessed using the point-biserial correlation coefficient between the metric values and functional states, as well as a modified version based on the difference in interquartile ranges of the metric values. We show that the most applicable metrics include power in the α, θ, and SMR frequency bands, as well as signal entropy in the 0.3–30 Hz range. These metrics demonstrate significant but variable changes between functional states for almost all experienced participants in the dataset.