RUFINA HUSSAIN , SHERIN ZAFAR , SAFDAR TANWEER
DOI: https://doi.org/Artificial intelligence (AI) holds significant potential to address mental health disparities, particularly in high-stress regions like the Gulf, where immigrant populations are vulnerable to psychological distress. This study presents a dual-modality AI framework designed to detect symptoms of stress, depression, and anxiety among immigrants in the United Arab Emirates, Saudi Arabia, Qatar, and Oman. Using a real-world dataset of DASS-21 responses, demographic profiles, personality traits (TIPI), and cognitive measures (VCL), a Random Forest classifier achieved 86.8% accuracy and a ROC-AUC of 0.94. In parallel, a convolutional neural network (CNN) was trained on publicly available brain MRI images to demonstrate the feasibility of integrating neuroimaging into mental health assessment. The framework incorporates explainable AI (XAI) techniques, including feature importance and correlation analysis, to ensure transparency and interpretability. Depression rates were notably higher among immigrants in the UAE and Oman, likely due to sociocultural and occupational stressors. The study emphasizes the ethical deployment of AI through fairness, privacy, and cultural sensitivity, offering a scalable and interpretable approach for mental health monitoring in underserved migrant communities.
