Interactive VR-based mobile training framework for adolescents with autism spectrum disorder

Virtual reality (VR) technologies have become powerful tools for delivering interactive learning experiences, offering controlled and adaptive environments for skill development. This study introduces an AI-enhanced standalone VR training framework designed for adolescents with autism spectrum disorder (ASD), integrating two interactive environments—a restaurant and a classroom—to support both life skills and educational competencies. The system was developed using Unity and Blender and delivered through Oculus Quest 3 standalone head-mounted displays with a dual-interface monitoring system for therapists and parents. A structured four-phase protocol (orientation, environment-specific training, integrated practice, and assessment) guided 15 participants (11 males, 4 females), aged 10–13 years, through 12 standardized sessions, producing 180 session-level records. Quantitative data included task completion time, error frequency, number of attempts, and interaction patterns, while qualitative feedback was collected from therapists and parents. Statistical analyses (paired t-tests, repeated measures ANOVA) revealed significant phase-related changes in completion time, error frequency, and task attempts, while success rates remained stable. An AI component using a Decision Tree classifier achieved 70.4% accuracy in predicting task outcomes, providing preliminary evidence for the role of machine learning in adaptive feedback and personalized interventions. Findings suggest that the proposed standalone VR framework enhances engagement and skill development among adolescents with ASD while offering valuable analytics for educators and therapists. The integration of VR, AI, and dual-environment design underscores the potential of immersive technologies to support scalable, adaptive, and data-driven interventions in special education.