Abstract
BACKGROUND: To compare the effects of various exercise modalities on intelligence and determine the optimal exercise dose for children and adolescents. METHODS: A systematic review and Bayesian network meta-analysis following PRISMA guidelines was conducted. Four databases were searched up to 1 April 2025. Eligible RCTs involved participants aged 5-18 years and assessed exercise interventions with intelligence outcomes (general, fluid, or crystallized). Standardized mean differences (SMDs) with corresponding 95% confidence interval or credible intervals were calculated. Dose-response relationships were analyzed using model-based network meta-analysis. RESULTS: Fifteen RCTs with 3,400 participants were included. Exercise was linked to small-to-moderate improvements in general (SMD = 0.59), fluid (SMD = 0.43), and crystallized intelligence (SMD = 0.64). Dual-task balance training (DTBT) produced the most consistent and significant benefits across all domains. Yoga and multi-component exercise also showed positive effects. Optimal outcomes were achieved with sessions lasting ≥117.7 min, three times weekly, totaling 220 min per week for at least 11.12 weeks. An inverted U-shaped dose-response curve indicated diminishing returns beyond the optimal frequency and duration. CONCLUSION: DTBT is the most effective exercise modality for improving intelligence in children and adolescents. The findings provide evidence-based guidance for designing school and clinical exercise programs to support cognitive development during growth.