Optimizing exercise prescriptions for older adults with mild cognitive impairment: a dose–response network meta-analysis of modality–dose combinations on executive function

Objective To use a Bayesian network dose–response meta-analysis to determine how overall exercise dose and different exercise modalities influence the three core domains of executive function-working memory, cognitive flexibility, and inhibitory control-in older adults with mild cognitive impairment. Methods PubMed, Embase, Web of Science, and the Cochrane Library were searched through 10 November 2025 for randomized controlled trials. Two reviewers independently screened eligible studies. Studies were included if they enrolled adults aged ≥60 years with mild cognitive impairment, evaluated repeated exercise programmes, included usual care, low-intensity activity, or another exercise modality as a comparator, and reported at least one outcome of working memory, cognitive flexibility, or inhibitory control. Studies were excluded if participants had dementia or other neurological conditions affecting cognition, if only acute single-bout exercise was examined, if the study was non-randomized or non-peer-reviewed, if essential outcome data were unavailable, or if the Physiotherapy Evidence Database score was <4. Exercise dose was standardized as metabolic equivalents (METs)⋅min/week, and Bayesian model-based network meta-analysis was used to estimate non-linear overall and modality-specific dose–response relationships. Results Thirty-eight studies met the eligibility criteria. Across all three executive-function dimensions, the overall dose–response curves exhibited a characteristic non-linear improvement pattern. Moderate exercise doses (approximately 300–1,000 METs⋅min/week) were associated with the most reliable gains, whereas higher-dose ranges showed plateaus or modest declines. Effective dose windows varied by modality and cognitive dimension: Exercise Joint Cognitive Training yielded the most consistent improvements in working memory, while Comprehensive Physical Training and Walking Aerobic Exercise demonstrated more stable effects on cognitive flexibility and inhibitory control. Sensitivity analyses supported the stability of the models. The certainty of evidence ranged from low to moderate, and publication bias was minimal. Conclusion Exercise yields a non-linear dose–response relationship with executive function in older adults with mild cognitive impairment, with moderate doses producing the most robust improvements. Distinct optimal dose windows exist across cognitive dimensions and exercise modalities, offering a basis for precision exercise–cognition prescriptions. These findings support prioritizing dose-informed intervention strategies in ageing and cognitive-decline management and highlight the need for future trials adopting multi-dose gradient designs to validate optimal dose ranges.