Depression screening using hybrid neural network.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Psychiatry

Background:

• Major depressive disorder (MDD) is a significant global health concern, with increased prevalence during the COVID-19 pandemic.
• Effective diagnostic tools are crucial for timely depression management.
• Machine learning and deep learning applied to electroencephalography (EEG) show potential for automatic depression detection.

Purpose of the Study:

• To develop and evaluate a machine learning model for accurate depression detection using multi-channel EEG signals.
• To compare the performance of a 2DCNN-LSTM classifier against traditional machine learning algorithms for EEG-based depression detection.

Main Methods:

• Utilized 128-channel EEG signals, applying simple filtering techniques.
• Employed a 2DCNN-LSTM classifier for depression detection.
• Conducted 24-fold leave-one-out cross-validation experiments.
• Compared the 2DCNN-LSTM model with Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Decision Tree (DT) algorithms.

Main Results:

• The 2DCNN-LSTM model achieved an average classification accuracy of 95.1% and an AUC of 0.98 for 6-second EEG signals.
• The proposed model significantly outperformed SVM (72.05%), KNN (79.7%), and DT (79.49%).
• 100% classification accuracy was achieved for 300-second EEG signal participants.

Conclusions:

• The 2DCNN-LSTM model demonstrates superior performance in detecting depression from EEG signals compared to traditional methods.
• This approach offers a highly accurate and efficient tool for depression diagnosis.
• The findings highlight the potential of deep learning with multi-channel EEG for advancing mental health diagnostics.