A Wearable EEG-Based Hybrid CNN-RNN Framework for Psychiatric Disorder Diagnosis: Leveraging XAI for Enhanced Clinical Insights

Mental health disorders like schizophrenia, anxiety, and depression are major causes of disability worldwide, yet diagnosis remains difficult due to the lack of objective biomarkers. This study introduces a hybrid CNN-RNN framework leveraging wearable EEG sensors and deep learning for more accurate diagnostics. CNNs extract spatial features, while RNNs model temporal patterns, improving performance over traditional and standalone deep learning models. Evaluated on public and custom datasets, the model achieves superior accuracy, precision, recall, F1-score, and ROC-AUC. Explainable AI techniques enhance interpretability, identifying clinically relevant EEG patterns. Cross-Dataset validation confirms robustness, though challenges like dataset size, real-time processing, and ethical concerns remain. This approach advances objective, data-driven mental health diagnostics, promoting more effective and accessible care.