EEG-based Major Depressive Disorder Detection Using Data Mining Techniques.

Area of Science:

• Neuroscience
• Psychiatry
• Biomedical Engineering

Background:

• Major Depressive Disorder (MDD) is a prevalent mental illness causing persistent low mood and despair.
• Current diagnostic methods rely on subjective rating scales, necessitating professional assessment.
• Objective biomarkers for MDD detection are needed to improve diagnostic accuracy and accessibility.

Purpose of the Study:

• To investigate the efficacy of electroencephalography (EEG) derived features in detecting Major Depressive Disorder (MDD).
• To compare the performance of logistic regression (LR) and support vector machine (SVM) algorithms using EEG-based beta-alpha-ratio features for MDD screening.
• To evaluate the superiority of data mining methods incorporating EEG features over traditional methods.

Main Methods:

• Recruited 92 participants for EEG signal collection at Shenzhen Traditional Chinese Medicine Hospital.
• Assessed MDD severity using the HAMD-17 rating scale administered by a trained physician.
• Employed logistic regression (LR) and support vector machine (SVM) with derived EEG-based beta-alpha-ratio features (LR-DF and SVM-DF) for MDD screening.

Main Results:

• The LR-DF method achieved an F1 score of 0.76 ± 0.30.
• The SVM-DF method achieved a superior F1 score of 0.92 ± 0.18.
• Both LR-DF and SVM-DF demonstrated significant advantages over LR and SVM methods without EEG-based features.

Conclusions:

• Derived EEG-based beta-alpha-ratio features, when analyzed with data mining techniques like SVM, show high potential for objective MDD detection.
• The SVM-DF approach offers a promising, data-driven alternative to subjective scale-based assessments for identifying patients with MDD.
• Further research into EEG signal analysis could lead to more accurate and accessible diagnostic tools for Major Depressive Disorder.