AI-DRIVEN MULTILINGUAL DISTRESS DETECTION AND PREDICTIVE WELLBEING ANALYTICS FOR COLLEGE STUDENTS

Abstract

This paper presents a simple, scalable AI system that supports the mental health of college students through real-time text-based interaction. The system uses a conversational agent with natural language processing to understand student messages, detect emotional distress, and classify risk levels as low, medium, or high. It supports multiple languages so that students can express themselves comfortably and anonymously. The AI then offers basic guidance, coping tips, and, in high‑risk situations, triggers alerts for human support. At the same time, anonymized data from many interactions is converted into predictive wellbeing analytics that show overall stress trends, emerging risk patterns, and engagement levels over time. These insights help institutions plan early interventions, allocate counselling resources, and design more supportive policies without exposing any student’s identity. The paper will describe the technical design, AI and NLP workflow, and how predictive modelling is used to move from reactive crisis response to proactive care in college settings.