Sustained attention detection in humans using a prefrontal theta-EEG rhythm

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning

Background:

• Sustained attention is vital for cognitive performance.
• Electroencephalography (EEG) rhythms offer insights into brain states.
• Differentiating attentive states using EEG requires robust analytical methods.

Purpose of the Study:

• To investigate the role of prefrontal theta-EEG rhythm in sustained attention monitoring.
• To classify high-focused and low-focused individuals based on EEG data.
• To evaluate the efficacy of machine learning classifiers in identifying attentive states.

Main Methods:

• EEG data (theta, alpha, beta, gamma rhythms) were collected from 20 participants performing mental tasks.
• Multi-stage discrete wavelet transform was used for EEG rhythm classification.
• K-Nearest Neighbour (KNN) classifier was trained on statistical features of EEG rhythms to distinguish attentive states.

Main Results:

• The KNN classifier achieved the highest f1-score of 88.88% for the theta-EEG rhythm.
• Alpha-EEG rhythm yielded an f1-score of 85.71%, while beta and gamma rhythms showed lower scores.
• Combining all EEG rhythms resulted in a lower f1-score (53.33%) compared to theta alone.

Conclusions:

• The theta-EEG rhythm is significantly more relevant than other EEG rhythms for identifying human attentive states.
• Machine learning, particularly KNN, can effectively utilize EEG features for attention state classification.
• Prefrontal theta-EEG activity serves as a reliable biomarker for sustained attention.