Adolescent Mental Health Assessment Based on Affective Computing Technology

Abstract

Adolescent mental health has become an important concern in schools and society, yet current assessment approaches still rely mainly on self-report questionnaires and clinician-administered interviews, which are limited in objectivity, scalability, and suitability for repeated monitoring. This study examined the feasibility of adolescent mental health assessment based on affective computing technology, with particular attention to its potential for student wellbeing assessment and early support in educational contexts. Using a cross-sectional diagnostic design, 383 adolescents aged 10–19 years were recruited from community and clinical settings in Shanghai, including 189 participants with clinically diagnosed major depressive disorder and 194 healthy controls. Participants completed a gamified “Whac-A-Mole” task on an online platform while facial videos were recorded by a 720p webcam at 16 frames per second. Forty visual behavioural features were extracted and used to train a Random Forest classifier. The model achieved an AUC of 0.813 on the internal test set and 0.886 on the external cohort. The findings support the feasibility of using affective computing and gamified interaction to develop more sensitive, scalable, and youth-friendly approaches to adolescent mental health assessment.