Abstract
BACKGROUND: Children with autism spectrum disorder (ASD) exhibit cognitive, motor, and social difficulties that affect engagement, causing developmental delays, behavioral challenges, and obesity-interrelated concerns in daily functioning and well-being. Although interactive interventions have incorporated physical activity, they often rely on limited physical involvement and lack iterative, expert-informed design, as built on pre-existing game frameworks. Physical activity is often operationalized as constrained input (eg, gestures or in-place actions) rather than exertion-intensive, whole-body exercise, and design guidance for adapting exercise content under ASD-oriented safety and cognitive-sensory constraints remains limited. These limitations highlight the need for exergames that promote sustained, full-body participation aligned with developmental goals, motivating formative, co-design with expertise and initial field testing in this population. OBJECTIVE: We aim to iteratively design exercise-based serious games (SGs) for children with ASD through a structured, expert-informed co-design process involving 21 professionals across special education, adapted physical education, and human-computer interaction, and to examine feasibility and use contexts through an exploratory multiple-case pilot study. METHODS: We derived serious exergames using 4 design methods-stakeholder interview, concept mapping, creative matrix, and visualize the vote. Two exergames-"Fruit Sorting Run" and "Hazard Avoiding Ride"-were developed, integrating full-body running and cycling movements into goal-directed tasks under ASD-oriented constraints. We conducted a multiple-case pilot with 3 children with ASD. During gameplay, caregivers labeled engagement using a binary input interface, and we conducted postsession caregiver interviews to capture complementary observations. RESULTS: Engagement in both exergames tended to increase over normalized time. Generalized estimating equations with a logit link and an autoregressive working correlation of order 1 (AR1), including participant indicators, showed a statistically significant association between normalized time and engagement in Fruit Sorting Run (per 0.1 increase: β=0.48; odds ratio 1.62, 95% CI 1.09-2.38; P=.02) and Hazard Avoiding Ride (per 0.1 increase: β=0.66; odds ratio 1.93, 95% CI 1.04-3.60, P=.04). Caregiver interviews reinforced these findings, reporting increased attention, motivation, and enjoyment across both activities. CONCLUSIONS: The findings support the applicability of an expert-informed design approach and the viability of the resulting exergames, integrating goal-directed physical activity, virtual agent-based prompting, and stakeholder-informed considerations such as motor-cognitive alignment, interactive scaffolding, and support for daily living skills. Distinct from prior SG approaches that operationalize physical activity through discrete gestures or in-place interactions, the proposed exergames embed sustained, exertion-intensive, whole-body movement within structured gameplay. Within this exploratory multiple-case pilot, engagement trajectories tended to increase over time. These preliminary observations provide an initial basis for a testable hypothesis that exertion-intensive, full-body SGs with virtual agent-based prompting may be associated with increasing engagement over time, meriting further examination in larger samples and applied educational and therapeutic contexts.