A Subject-Specific Attention Index Based on the Weighted Spectral Power

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Attention is a critical cognitive function.
• Electroencephalography (EEG) offers convenient, real-time attention monitoring.
• Existing EEG-based attention indices have limitations in reliability due to inter-subject variability and empirical frequency bands.

Purpose of the Study:

• To develop a subject-specific attention index using weighted spectral power from EEG.
• To improve the accuracy and reliability of attention monitoring compared to traditional methods.
• To overcome the limitations of empirical frequency bands and inter-subject variance.

Main Methods:

• Proposed a subject-specific attention index based on weighted spectral power of EEG.
• Determined frequency band ranges from subject-specific spectral power changes, not empirical values.
• Utilized the ratio of power spectral density (PSD) in attentional vs. inattentional states to calculate weights.
• Evaluated the index on two open EEG datasets (44 subjects) and compared it with three traditional indices.

Main Results:

• The proposed subject-specific attention index demonstrated higher accuracy in describing attentional states.
• Achieved accuracies of 86.21% (t-test) and 70.00% (Wilcoxon rank-sum test) at the 1% significance level.
• Showed significant improvements of 41.38% and 20.00% over the best traditional indices.
• Indicated reduced inter-subject variance and enhanced effectiveness through weighted spectral power.

Conclusions:

• The developed subject-specific attention index offers a more accurate and reliable method for attention monitoring.
• Weighted spectral power analysis effectively enhances EEG-based attention assessment.
• This approach provides an efficient tool for real-time measurement of attentional states.