Electroencephalogram-Based Sustained Attention Assessment Using Sparse Model for Feature Selection.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Biomedical Engineering

Background:

• Sustained attention (concentration) is crucial for task performance.
• Traditional concentration assessment relies on subjective reports and behavioral measures.
• Electroencephalography (EEG) presents a promising neurophysiological method for objective attention quantification.

Purpose of the Study:

• To investigate the neural correlates of concentration using EEG.
• To identify optimal EEG spatial and spectral features for concentration assessment.
• To determine if concentration can be reliably assessed with a reduced set of EEG features.

Main Methods:

• A go/no-go task was utilized while recording EEG signals and behavioral response times.
• Group Lasso feature selection was applied to a wide array of EEG parameters (spectral powers, ratios, entropy).
• Identification of the most predictive combinations of spatial and spectral EEG features.

Main Results:

• Frontal, right parietal, and occipital electrodes, along with their power ratios, emerged as the most effective variables for concentration assessment.
• A small subset of electrodes and features was sufficient for accurate concentration evaluation.
• Reduced feature sets did not significantly impact overall performance metrics.

Conclusions:

• EEG, particularly specific frontal, parietal, and occipital regions, provides robust neurophysiological markers for sustained attention.
• Concentration assessment can be achieved efficiently using a parsimonious selection of EEG features and electrodes.
• This neurophysiological approach offers a more objective and potentially more sensitive method for evaluating attentional states.