Effects of mind-body training on upper-limb function in stroke patients: a multilevel dose-response meta-analysis

Le résumé fourni par la source

Objective: This systematic review and meta-analysis examined the effects of mind-body training (MBT) on upper-limb function in patients with stroke. A Bayesian dose-response model and an XGBoost-SHAP analysis were used to assess the optimal intervention dosage and to determine key factors influencing outcomes. Methods: PubMed, Web of Science, PsycINFO, and the Cochrane Library were searched from their inception to January 10, 2026, for randomized controlled trials. Effect sizes were calculated as standardized mean differences (SMDs). Risk of bias was assessed using ROB2. Analyses included Egger's test, trim-and-fill, sensitivity analysis, subgroup analysis, meta-regression, trial sequential analysis, and Bayesian hierarchical dose-response modeling. XGBoost-SHAP was used for exploratory analysis of predictors. Results: = 0.753). GRADE assessment showed moderate-certainty evidence. Exploratory subgroup analyses suggested larger effects in studies involving subacute patients (SMD = 0.95) and Tai Chi interventions (SMD = 0.99). Nonlinear dose-response analysis revealed an inverted U-shaped relationship, with an optimal cumulative duration of about 33.5 h (range: 32-35 h), corresponding to 5 sessions per week × 30 min × 14-15 weeks. Exploratory XGBoost-SHAP analysis indicated that patient type had the highest relative importance among the study-level predictors included. Conclusion: Moderate-certainty evidence indicates that MBT effectively enhances upper-limb function in stroke patients, with the best results seen at around 33.5 total h. A schedule of 30-min sessions, five times a week for 14-15 weeks, may provide preliminary guidance for future dose-stratified trials and clinical planning, particularly in studies involving subacute-stage patients, although this finding should be confirmed in future stage-stratified RCTs. Systematic review registration: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261334249, identifier: CRD420261334249.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.