AI-Driven Eye-Tracking and Emotion Recognition to Improve Attention Across Storytelling Modalities in Children with Autism Spectrum Disorder

Children with Autism Spectrum Disorder (ASD) often face challenges in attention regulation, information processing, and emotional engagement, affecting learning outcomes. This study investigates an AI-driven interactive storytelling system integrating webcam-based eye tracking and facial emotion recognition to assess visual attention, emotional engagement, and comprehension. Thirty children with ASD evaluated three storytelling formats: static images, dynamic animations, and interactive presentations. Eye-tracking metrics (fixation duration, time to first fixation, gaze transition frequency) quantified attention, while emotion recognition evaluated affective responses. Results revealed distinct engagement patterns. The interactive format produced longer fixation durations, higher gaze transitions, and the highest comprehension accuracy (83.2%), compared to dynamic (72.1%) and static (61.8%) formats. Dynamic presentations elicited faster initial visual attention. Post-test results showed improved comprehension following storytelling exposure. These findings suggest AI-driven interactive storytelling may enhance engagement and learning in children with ASD under controlled conditions.