A comprehensive exploration of machine learning techniques for EEG-based anxiety detection.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomedical Engineering

Background:

• Electroencephalogram (EEG)-based systems performance relies heavily on feature extraction and machine learning algorithms.
• Accurate anxiety detection using EEG is vital for mental health monitoring and intervention.

Purpose of the Study:

• To investigate the impact of different labeling methods, feature extraction techniques, and machine learning algorithms on EEG-based anxiety detection accuracy.
• To identify the optimal combination of algorithms for enhanced performance in detecting anxiety states.

Main Methods:

• Utilized Hamilton Anxiety Rating Scale (HAM-A) and Self-Assessment Manikin (SAM) for anxiety state labeling.
• Employed Discrete Wavelet Transform (DWT) and Power Spectral Density (PSD) for EEG feature extraction.
• Compared ensemble methods (Random Forest, AdaBoost, Gradient Bagging) against conventional classifiers (LDA, SVM, KNN).

Main Results:

• Hamilton Anxiety Rating Scale (HAM-A) labeling combined with Discrete Wavelet Transform (DWT) features consistently provided superior results across all tested classifiers.
• The Random Forest (RF) classifier achieved the highest accuracy (87.5%), demonstrating superior performance in accuracy, precision, and recall.
• AdaBoost bagging classifier followed with 79% accuracy.

Conclusions:

• The combination of HAM-A labeling and DWT feature extraction is highly effective for EEG-based anxiety detection.
• Random Forest (RF) emerges as the most accurate and robust classifier for this application.
• Optimizing algorithm selection is critical for advancing the efficacy of EEG-based mental health diagnostic tools.