AI-Based Remote Assessment of Depression in Humans: “A Pathway to Enhancing Food and Job Security for Poverty Reduction in Nigeria."

Abstract

Depression remains a significant public health concern in Nigeria, exacerbated by limited mental health services, economic instability, and food insecurity.Early detection is critical for intervention, but existing methods are inaccessible, expensive, and stigmatised.This study proposes an AI-driven, multimodal depression assessment model that integrates textbased sentiment analysis, voice tone recognition, and facial expression analysis, secured with blockchain technology for data privacy and trust.The model was developed using BERT for text analysis, SVM for voice classification, and CNN for facial emotion detection.Performance evaluation was based on accuracy, precision, recall, F1-score, and ROC-AUC.Results showed an accuracy of 95%, precision of 93%, recall of 96%, and F1-score of 94% over 20 training epochs.The ROC-AUC score reached 0.80, indicating strong classification performance in distinguishing depressed and non-depressed individuals.This research is significant as it introduces a scalable, AI-powered mental health assessment framework tailored to Nigeria's unique challenges, including rural inaccessibility and stigma.By automating depression screening, this model offers early intervention, reduces job losses, and promotes economic stability, with potential applications in telemedicine and mental health policy-making.This study demonstrates the feasibility and effectiveness of AIdriven depression detection, showing that a multimodal approach enhances classification accuracy.The integration of blockchain technology ensures secure and trustworthy mental health assessments, paving the way for wider adoption of AI in mental healthcare.