An Intelligent Exercise Intervention for Enhancing Hot Executive Functions in Children With...

Deficits in hot executive functions (EFs), including risky decision-making, delay discounting, and emotion regulation, represent important yet relatively understudied features of attention-deficit/hyperactivity disorder (ADHD). Although physical exercise has been shown to improve cool EFs, its effects on hot EFs remain uncertain. Furthermore, conventional intervention approaches are often limited by accessibility and implementation barriers, underscoring the need for intelligent, home-based alternatives.

In this three-arm randomized controlled trial, 145 children aged 6-10 years with ADHD were randomly assigned to an intelligent exercise group (EG1; n = 53), a traditional offline exercise group (EG2; n = 45), or a wait-list control group (CG; n = 47). Both intervention groups participated in a 12-week program consisting of three sessions per week, each lasting 45-60 minutes and integrating physical and cognitive training. Participants in EG1 completed the intervention through a custom-developed WeChat mini-program ("LeDongYing"), which incorporates computer vision-based motor assessment, dynamically personalized exercise prescriptions, real-time feedback, an artificial intelligence (AI)-assisted guidance agent, and a tailored exercise library. Participants in EG2 received coach-led, face-to-face training sessions. Outcome measures included risky decision-making, delay discounting, and emotion regulation. Intervention effects were evaluated using generalized linear mixed models, with potential confounders included as covariates.

For risky decision-making, EG1 demonstrated significantly greater improvements than CG in Block 3 (β = 3.38, p = 0.019), Block 4 (β = 6.89, p < 0.001), and total net score (β = 13.07, p = 0.012). EG2 also showed significantly greater improvements than CG in Block 3 (β = 3.49, p = 0.019) and Block 4 (β = 4.38, p = 0.009). No significant differences were observed between EG1 and EG2 for any risky decision-making outcome (all p > 0.05). For delay discounting, both EG1 (β = 0.12, p < 0.001) and EG2 (β = 0.12, p < 0.001) demonstrated significantly greater improvements than CG, with no significant difference between the two intervention groups (β = 0.00, p = 0.968).For emotion regulation, both EG1 (β = 3.38, p = 0.002) and EG2 (β = 2.71, p = 0.014) improved significantly more than CG, whereas no significant difference was observed between EG1 and EG2 (β = 0.66, p = 0.507). Adherence and participant satisfaction did not differ significantly between EG1 and EG2 (both p > 0.05).

The intelligent exercise intervention appears to be an effective, scalable, and accessible non-pharmacological approach for improving hot executive functions in children with ADHD. Its efficacy was comparable to that of traditional face-to-face exercise programs, suggesting that this delivery model may help overcome implementation barriers and expand access to rehabilitation services without implying superiority over conventional approaches.

The study has been registered on https://www.chictr.org.cn/ (registration number: ChiCTR2200065413; registration link: https://www.chictr.org.cn/showproj.html?proj=182412).