AI-DASA: AI-Based Depression, Anxiety, and Stress Assessment

Minaya W.; Aramburu F.; Santisteban J. · Engineering, Technology and Applied Science Research (2025) · DOI: 10.48084/etasr.13210

Resumen

This article presents the design, implementation, and validation of AI-DASA, an innovative web-based predictive system designed to detect emotional disorders, specifically anxiety, depression, and stress, in university students at an early stage. AI-DASA is based on advanced Natural Language Processing (NLP) techniques and utilizes large-scale language models to analyze free-text documents written by students. The system was evaluated in a three-week experiment involving expert psychologists and 200 university students who interacted with the platform. Validation was conducted at three levels: operational efficiency compared to the traditional process, usability from the perspective of mental health experts, and overall satisfaction perception among students. The results show that AI-DASA achieved an average accuracy of 94% in detecting disorders, significantly reducing the evaluation time from 48 hours to just 20 minutes per case and eliminating the need for direct human intervention in the initial diagnostic phase. Both experts and students reported high levels of satisfaction and usability. This system represents a promising tool for emotional screening in educational settings, particularly in environments with limited access to mental health professionals, thus contributing to the earlier and more efficient detection of emotional problems in the student population. © (2025), (Dr D. Pylarinos). All right reserved.

Instituciones: Universidad Peruana de Ciencias Aplicadas