Optimal type and dose of exercise to improve cognitive function in healthy and pre-sarcopenic older adults: a bayesian network meta-analysis of randomized controlled trials

BACKGROUND: Exercise is increasingly recognized as a non-pharmacological strategy for cognitive aging; however, comparative evidence across modalities, phenotypes, and doses is limited. METHODS: We conducted a Bayesian network meta-analysis of 38 randomized controlled trials (N = 4,047; 88 arms). The interventions included aerobic, resistance, multimodal, and other exercise formats compared with non-exercise controls. The primary outcome was global cognition. Analyses were stratified by age (< 70 vs. ≥70 years) and phenotype (healthy vs. pre-sarcopenic). Dose-response relationships were modeled using the weekly volume (MET·min/week). RESULTS: Aerobic (standardized mean difference [SMD] 0.58, 95% CI 0.33-0.83), resistance (0.62, 0.35-0.88), and multimodal programs (0.68, 0.40-0.95) significantly improved cognition compared to the control, with smaller effect sizes. Healthy older adults benefited most from aerobic (0.88, 0.55-1.20) and resistance training (0.80, 0.42-1.19), whereas multimodal programs were most effective for pre-sarcopenia (0.60, 0.29-0.90). Dose-response analysis showed clinically meaningful benefits from ~ 600 MET·min/week, with optimal effects between 700 and 1,200 MET·min/week. Higher volumes conferred no consistent additional gains in pre-sarcopenia. CONCLUSIONS: Exercise is a scalable, safe, and clinically effective approach for preserving late-life cognition. For healthy older adults, aerobic or resistance training at ≥ 600 MET·min/week is recommended; for pre-sarcopenic individuals, multimodal programs at approximately 700-800 MET·min/week offer the best balance of efficacy and sustainability. These findings provide actionable targets for clinicians and policymakers, advancing precision exercise prescriptions for cognitive health in aging populations.