AI system developed by 'Alexandru Ioan Cuza' University can detect depression from social media posts

A new AI program from Alexandru Ioan Cuza University analyzes posts on X for signs of depression and provides explanations for its assessments. It considers both the content and timing of posts, aiming to enhance transparency and trust. The system, developed by Associate Professor Madalina Raschip, uses a Transformer-based model to interpret text and generate understandable reasons for its conclusions. Trained on 20,000 posts, it achieved an 85% accuracy rate in identifying depression-related content. The research highlights the correlation between posting times and mental states, with findings indicating that depressive posts are more common at night.