An Interactive System for Recognizing and Enhancing Engagement in Children with Autism Spectrum Disorder

Abstract

Autism spectrum disorder (ASD) is associated with difficulties in cognitive, motor, and social functioning, which can limit children’s opportunities for participation and skill development. Active engagement plays a critical role in supporting developmental progress, yet many existing intervention systems require complex technologies or lack adaptability to children’s varying needs. This dissertation explores interactive approaches to support engagement in children with ASD across three studies. The first study focuses on the design and early evaluation of serious exergames, developed through a structured co-design process with special educators, therapists, and researchers. The second study presents the Engagnition dataset, comprising behavioral and physiological signals collected during physical activity–based gameplay. It further introduces deep learning models for engagement recognition and intervention estimation. The third study proposes InteractiSense, a prototype for estimating social engagement during collaborative play. This prototype was evaluated through user studies involving children with ASD and follow-up interviews with caregivers. Together, these studies examine how structured game design, multi-modal sensing, and adaptive systems can be used to support and measure engagement in developmentally relevant contexts. By combining expert-informed design processes, data-driven modeling, and early-stage system evaluation, the dissertation contributes to the design of accessible and responsive technologies for children with ASD.