Automated Depression Detection Using Deep Representation and Sequence Learning with EEG Signals.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Depression is a prevalent global mood disorder.
• Manual analysis of electroencephalogram (EEG) signals for depression diagnosis is time-consuming and requires expertise.
• Automated systems are needed to assist clinicians in diagnosing depression.

Purpose of the Study:

• To develop and evaluate a deep hybrid model for automated depression detection using EEG signals.
• To leverage Convolutional Neural Network (CNN) and Long-Short Term Memory (LSTM) architectures for enhanced signal analysis.

Main Methods:

• A deep hybrid model combining CNN and LSTM was developed.
• CNN layers were used to learn temporal properties of EEG signals.
• LSTM layers were employed for sequence learning from EEG data.
• EEG signals from both left and right brain hemispheres were utilized.

Main Results:

• The CNN-LSTM model achieved 99.12% classification accuracy for right hemisphere EEG signals.
• The model achieved 97.66% classification accuracy for left hemisphere EEG signals.
• The developed model demonstrates high accuracy and speed in depression detection.

Conclusions:

• The proposed CNN-LSTM model is an accurate and efficient tool for detecting depression using EEG signals.
• This automated system can aid psychiatrists in clinical settings, particularly in psychiatry wards.
• The findings support the integration of AI-driven diagnostic tools in mental healthcare.