EEG-based Depression Detection Using Convolutional Neural Network with Demographic Attention Mechanism.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Machine Learning

Background:

• Electroencephalography (EEG) is a key tool for depression detection, but signal complexity and individual differences pose challenges.
• Existing algorithms struggle with generalization due to demographic variations like age and gender.
• Incorporating demographic data can refine EEG-based depression detection models.

Purpose of the Study:

• To develop an improved EEG-based depression detection method by integrating demographic factors.
• To enhance the feature extraction capabilities of one-dimensional Convolutional Neural Networks (1-D CNNs) using demographic data.
• To investigate the effectiveness of an attention mechanism for combining EEG signals with gender and age information.

Main Methods:

• A one-dimensional Convolutional Neural Network (1-D CNN) was designed for EEG signal feature extraction.
• An attention mechanism was employed to integrate gender and age data into the 1-D CNN architecture.
• The model was trained and validated on a dataset of 170 subjects (81 depressed, 89 controls).

Main Results:

• The proposed method, incorporating demographic factors via an attention mechanism, outperformed a standard 1-D CNN.
• The integrated approach showed superiority over other methods of incorporating demographic data.
• The model effectively captured complex correlations between EEG signals and demographic factors.

Conclusions:

• The organic integration of EEG signals and demographic factors offers a promising avenue for accurate depression detection.
• This approach enhances the reliability and generalization of AI-driven diagnostic tools in mental health.
• The study highlights the clinical relevance of personalized, data-driven approaches in psychiatric diagnostics.