Optimal exercise dose and training modalities for improving gross motor function in children with cerebral palsy: a systematic review and Bayesian network meta-analysis of RCTs.

UNLABELLED: The objective of this study is to clarify the comparative efficacy of six distinct exercise modalities in improving domain-specific gross motor functions among children with cerebral palsy (CP) and to identify the specific intervention that could produce optimal functional gains and to investigate the dose-response relationship between exercise exposure and motor outcomes, thus determining evidence-informed optimal dose ranges across key functional domains. We conducted a systematic review and meta-analysis following the inclusion of randomized controlled trials (RCTs) published up to March 2026 across four electronic databases. Pairwise meta-analyses and network meta-analyses were performed using Stata version 15.1. Dose-response relationships were modeled using restricted cubic splines in R version 4.0.3. Twenty-nine RCTs involving 886 children with CP met inclusion criteria. Pairwise meta-analyses revealed that robot-assisted gait training conferred a statistically significant advantage over comparator interventions in improving walking, running, and jumping skills using Gross Motor Function Measure-88 dimension E (SMD = 0.48, 95% CI 0.02-0.93, P = 0.042). In contrast, resistance/strength training demonstrated the largest effect on balance function (SMD = 0.95, 95% CI 0.38-1.51, P = 0.001). Network meta-analysis ranked interventions using surface under the cumulative ranking curve values, with clinical interpretation anchored to statistically significant pairwise contrasts. Dose-response analyses indicated a nonlinear association between exercise dosage and functional improvement-optimal dose ranges varied meaningfully across motor domains, supporting domain-specific dosing strategies for personalized rehabilitation. CONCLUSION: Different exercise modalities appear to confer differential benefits across gross motor domains, though evidence certainty is rated as "very low." The observed nonlinearity in dose-response curves suggests that both subtherapeutic and supraoptimal dosages may attenuate treatment effect. However, these findings remain exploratory and require confirmation in larger, methodologically rigorous RCTs before translation into clinical guidelines or practice. WHAT IS KNOWN: • Exercise improves gross motor function in children with cerebral palsy, but evidence on optimal exercise modality and dose remains limited and inconsistent. WHAT IS NEW: • Robot-assisted gait training significantly improved standing and walking/running/jumping abilities, whereas resistance training conferred the largest benefit on balance. • Nonlinear dose-response relationships were identified, with domain-specific optimal ranges (e.g., ~ 248 METs·hours for walking/running/jumping).