The protocol for this meta-analysis was prospectively registered on PROSPERO (International Prospective Register of Systematic Reviews; registration number: CRD420251162082). The design, conduct, and reporting of this review adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 statement (Page et al., 2021) to ensure a rigorous and transparent methodology in literature search, study selection, data extraction, risk of bias assessment, and data synthesis, thereby enhancing the reproducibility and reliability of the findings.

A comprehensive literature search was conducted across five electronic databases: PubMed, Embase, Web of Science, Cochrane Library, and CNKI (China National Knowledge Infrastructure), covering records from their inception to October 25, 2025. Studies published in English and Chinese were included to ensure broad coverage of relevant research. The search strategy was constructed by combining Medical Subject Headings (MeSH) terms and free-text words related to three key concepts: (1) attention deficit hyperactivity disorder (ADHD), (2) exercise or physical activity interventions, and (3) mental health (MeSH, 2025). Boolean logical operators (AND, OR) were used to appropriately combine these components (MeSH, 2025).

The inclusion criteria were established based on the PICOS (Participants, Interventions, Comparisons, Outcomes, Study Design) framework (Amir-Behghadami and Janati, 2020). Participants (P): Children and adolescents (under 18 years of age) with a formal diagnosis of ADHD, irrespective of gender, region, or language. Diagnosis could be based on DSM, ICD, or other recognized diagnostic criteria. Interventions (I): Any intervention primarily based on physical activity or exercise, including sports, fitness training, mind-body exercises (e.g., yoga, tai chi), and virtual reality-based exercises. There were no restrictions on the specific type, intensity, frequency, or duration. Comparisons (C): Control groups receiving no intervention, usual care, or other non-exercise interventions. Multi-arm studies were eligible if the exercise interventions differed significantly between groups in type, intensity, or modality. Outcomes (O): At least one objectively measured, standardized, and quantifiable mental health outcome, such as scores from validated scales assessing emotion regulation, depression, or anxiety. Study Design (S): Randomized controlled trials (RCTs) or quasi-RCTs that reported sufficient statistical data (e.g., means, standard deviations, sample sizes) for effect size calculation.

Studies were excluded if they: (1) involved mixed populations without separately reported data for the ADHD subgroup; (2) included participants outside the child/adolescent age range without subgroup data; (3) had an intervention where the exercise component was not clearly identifiable or poorly described; (4) compared intervention and control groups where the exercise nature was not substantially different (e.g., differing only in delivery setting or personnel); (5) lacked objectively measured, standardized, and quantifiable mental health outcomes; (6) lacked sufficient statistical data even after attempting to contact the authors; or (7) were reviews, commentaries, case reports, conference abstracts, or non-peer-reviewed publications.

Two reviewers (ZSN, SWY) independently screened the titles and abstracts of references retrieved from the databases. They then assessed the full texts of potentially eligible studies, recording their decisions (0 = exclude/1 = include for full-text review) in a pre-formatted Excel spreadsheet. After removing duplicates, the reviewers independently screened the titles and abstracts of the remaining records, excluding those clearly ineligible. The full texts of the remaining articles were obtained and thoroughly evaluated. If a study was excluded, the reason was documented. If reported information was insufficient, the corresponding author was contacted via email, allowing a 2-week response period. Studies remained excluded if necessary data could not be obtained after these efforts. Any disagreements between the reviewers were resolved through discussion or by consulting a third reviewer (HYQ). The detailed selection process is illustrated in the PRISMA flow diagram (Figure 1).

Two assessors (ZSN, SWY) independently evaluated the methodological quality of the included studies using the PEDro scale and assessed the risk of bias using the Cochrane Risk of Bias tool for randomized trials (ROB 2) (Sterne et al., 2019). The PEDro scale is an 11-item checklist designed to evaluate the methodological rigor and statistical reporting of clinical trials (the ‘eligibility criteria’ item is not scored). Each satisfied item scores 1 point. Total scores were used to categorize studies as poor (<4), fair (4–5), good (6–8), or excellent (9–10) quality. The ROB 2 tool assesses bias across five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Studies were classified as having “low risk,” “some concerns,” or “high risk” of bias.

To maximize accuracy and consistency, data extraction was performed independently by two researchers. Subsequently, all extracted data underwent cross-checking and verification by another team member as a further quality control measure. The extracted information included: (1) study design and basic characteristics; (2) participant demographics and clinical features (e.g., age, gender distribution); (3) detailed intervention parameters (type, frequency, intensity, duration); (4) details of the control group; (5) outcome measurement methods (Follmann et al., 1992), assessment tools, time points, and statistics necessary for effect size calculation (means, standard deviations, sample sizes). All data were entered into a structured Excel database for subsequent analysis (Drevon et al., 2017). If key data were missing, systematic attempts were made to contact the corresponding authors via email, with a reminder sent after 48 h. Studies were excluded from the final analysis if the necessary data remained unavailable.

Statistical analyses were performed primarily using Stata 18.0 software, encompassing effect size synthesis, subgroup analysis, sensitivity analysis, meta-regression, and publication bias assessment. To combine results from studies using different measurement tools, the effect size was expressed as the Standardized Mean Difference (SMD) with a 95% confidence interval (CI). The absolute values of SMD were interpreted as follows: < 0.2 (negligible), 0.2 to < 0.5 (small), 0.5 to < 0.8 (medium), and ≥ 0.8 (large).

Between-study heterogeneity was quantified using the I² statistic, interpreted as: I² < 25% (low), 25% ≤ I² < 50% (moderate), 50% ≤ I² < 75% (substantial), and I² ≥ 75% (considerable). A fixed-effect model was used if heterogeneity was not significant (I² < 50% and p > 0.10); otherwise, a random-effects model was applied. Sensitivity analyses were conducted to test the robustness of the pooled results when I² ≥ 50%. Publication bias was assessed through visual inspection of funnel plots and Egger’s regression test. If significant publication bias was indicated, the “trim and fill” method was employed to adjust the effect estimate. To ensure adequate power, publication bias analysis was performed only when at least five studies were available for an outcome.

Since the included studies used various scales to assess depression, emotion regulation, and anxietyas performed only when at least dard deviations, sample sizes). All data were entered inoint. xcluding those clearly ineligible. The full tof effect sizes. For studies where a lower score indicated improvement, the mean and standard deviation were multiplied by -1. This adjustment ensured that for all studies, a higher score consistently represented a better outcome, simplifying the interpretation of the overall analysis.

Pre-specified subgroup analyses and meta-regressions were conducted to explore potential sources of heterogeneity and examine the influence of study-level moderators on the intervention effects, where data permitted. Subgroup analyses were performed for the outcomes of depression, anxiety, and emotion regulation based on the following a priori variables: (1) Exercise intensity, categorized as low (LPA, e.g., yoga, tai chi), moderate (MPA, e.g., brisk walking, jogging, typically 50–70% of maximum heart rate), or moderate-to-vigorous (MVPA, e.g., high-intensity interval training); (2) Exercise type, categorized as physical-dominant (e.g., running, swimming) or mind-body integrated (e.g., yoga, tai chi); (3) Age group, dichotomized as children ( ≤ 12 years) or adolescents ( > 12 years) based on mean age; (4) Intervention duration, categorized as a single session, ≤ 8 weeks, or > 8 weeks.

Meta-regression analysis was planned provided a sufficient number of studies (typically ≥ 10) were available to ensure statistical power (Sánchez-Meca and Botella, 2024). A random-effects meta-regression model using the restricted maximum likelihood method was employed. In this study, only the depression outcome met the minimum sample size requirement for meta-regression. Therefore, meta-regression was conducted solely for depression to explore the relationship between continuous or ordinal variables (e.g., age, intervention duration) and the effect size, with results presented as the regression coefficient (β), standard error (SE), and corresponding p-value.

The literature screening and selection process followed the PRISMA statement guidelines. Systematic searches across five databasesches acron = 1,352), Web of Science (n = 854), Embase (n = 780), Cochrane Library (n = 756), and CNKI (n = 79)), and CNKItotal of 3,821 initial records. After removing 1,224 duplicate records, we screened the titles and abstracts of the remaining 2,597 articles, excluding 2,127 that did not meet the inclusion criteria. Full-text articles were sought for the remaining 470 citations, of which 20 were unavailable. A detailed assessment of the 450 successfully retrieved full-text articles led to the exclusion of 428 studies. The primary reasons for exclusion were: lack of separate data for an ADHD population, unclear intervention description, no substantial difference between the intervention and control groups, absence of standardized emotional outcome measures, and insufficient statistical data. Ultimately, 22 studies met the inclusion criteria for the qualitative synthesis, and 18 of these, which fulfilled all criteria for meta-analysis, were included in the quantitative synthesis. All included studies were randomized or quasi-randomized controlled trials evaluating the effect of exercise interventions on mental health in children and adolescents with ADHD. The entire screening process was conducted independently by two researchers, with any disagreements resolved by a third researcher. The detailed flow of the literature screening process is depicted in Figure 1.

A total of 22 empirical studies from 12 countries—namely Egypt, Israel, Canada, China, Spain, Tunisia, the United States, Australia, Germany, South Korea, Iran, and Brazil—were included in the final qualitative synthesis. Participants in all studies were diagnosed with Attention-Deficit/Hyperactivity Disorder (ADHD) according to international standards such as the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV to DSM-5) or the International Classification of Diseases (ICD-10), ensuring diagnostic standardization and comparability across the sample.

The exercise interventions were diverse, primarily categorized into physical-dominant exercises (e.g., structured physical activities) and mind-body integrated exercises (e.g., yoga, tai chi). The intervention parameters varied widely: duration ranged from 1 to 20 weeks, frequency from 1 to 7 sessions per week, session length from 10 to 90 min, and intensity spanned low, moderate, and high levels, reflecting the heterogeneity of exercise prescriptions.

Control conditions included a variety of designs such as treatment-as-usual, no intervention, sedentary activities, cognitive training, and pharmacotherapy, providing effective comparators for assessing the additive effect of exercise interventions.

Outcomes were assessed using standardized and validated tools, including the Conners’ Rating Scales (CRS), Child Behavior Checklist (CBCL), and Strengths and Difficulties Questionnaire (SDQ), which quantified effects on emotional symptoms and emotion regulation, thereby ensuring the reliability of the findings (for detailed results, see Table 1).

According to the pre-established protocol, we used the revised Cochrane Risk of Bias tool (ROB 2) to assess the risk of bias in the 18 included randomized controlled trials (Sterne et al., 2019). Overall, exercise intervention studies primarily faced challenges in terms of intervention adherence, implementation of blinding procedures, and allocation concealment, reflecting the inherent methodological limitations of such research. However, most studies demonstrated a low risk of bias concerning the generation of random sequences and the selective reporting of outcome data (for details, please refer to Figures 2, 3).

Simultaneously, we employed the PEDro scale to quantitatively evaluate the overall methodological quality of the included studies PEDro (2025). The PEDro scale is a widely used tool for assessing the quality of randomized controlled trials (RCTs), encompassing 11 key items related to study design, allocation methods, blinding, follow-up, and outcome reporting. Each item is scored as “yes” or “no,” with the total score reflecting the study’s overall quality. The total score ranges from 0 to 10, where a higher score indicates stronger internal validity and more reliable evidence. In this study, PEDro scores ranged from 5/10 to 8/10, indicating that the included studies generally possessed moderate to good methodological quality, although limitations such as incomplete implementation of blinding and insufficient reporting of allocation concealment information were noted (for detailed results, see Table 2).

Among the 18 studies included in the meta-analysis, 11, 8, and 7 studies reported outcome measures for depression, emotion regulation, and anxiety, respectively. As the assessment tools varied across studies (including a total of 10 different scales such as CBCL, CRS, and BDI), the standardized mean difference (SMD) was used as the pooled effect size. Heterogeneity tests indicated the presence of heterogeneity across all outcome dimensions: depression (I² = 48.6%), emotion regulation (I² = 52.0%), and anxiety (I² = 76.9%). Although the I² value for depression was slightly below 50%, its heterogeneity test P-value was < 0.05. Consequently, a random-effects model was employed to pool the effect sizes for all analyses. The meta-analysis revealed that exercise intervention demonstrated significant beneficial effects on ADHD children and adolescents, showing improvement in depressive symptoms (SMD = 0.42, 95% CI: 0.129–0.712, P < 0.05), enhancement of emotion regulation capacity (SMD = 0.47, 95% CI: 0.06–0.924, P < 0.05), and reduction of anxiety symptoms (SMD = 0.43, 95% CI: 0.006–0.84, P < 0.05). All these differences were statistically significant (for detailed results, see Figures 4–6).

This study systematically evaluated the influence of potential moderating variables—including exercise intensity, exercise type, age group, and intervention duration—on the effects of exercise on mental health outcomes (depression, emotion regulation, and anxiety) through subgroup analyses.

Regarding exercise intensity, subgroup analyses revealed that only moderate-intensity exercise yielded a statistically significant improvement in depressive symptoms (SMD = 0.54, 95% CI: 0.18–0.90, P = 0.003). The effects of low-intensity and moderate-to-vigorous intensity exercise on depression, as well as the effects of all intensity levels on emotion regulation and anxiety, were non-significant (all P > 0.05). In the comparison of exercise types, mind-body exercises (e.g., yoga, tai chi) demonstrated significant beneficial effects on both depression (SMD = 0.46, 95% CI: 0.15–0.78, P = 0.004) and anxiety (SMD = 0.80, 95% CI: 0.56–1.61, P = 0.002), with a particularly large effect size observed for anxiety. In contrast, purely physical exercises did not show significant effects on any of the three outcome measures (all P > 0.05). From the perspective of age groups, exercise interventions produced a significant improvement in depressive symptoms among adolescents (SMD = 0.35, 95% CI: 0.01–0.70, P = 0.046). Although a moderate effect trend was observed for children, it did not reach statistical significance (P = 0.059). Exercise interventions did not show significant effects on emotion regulation or anxiety in either age group. Concerning intervention duration, no statistically significant differences were found in the improvement of depression, emotion regulation, or anxiety across different intervention lengths (single session, ≤ 8 weeks, > 8 weeks; all P > 0.05) (for detailed results, see Table 3).

To further explore potential linear relationships with continuous moderators (age and intervention duration), simple meta-regression analyses were performed. The results indicated that neither age (β = -0.20, SE = 0.32, P = 0.552) nor intervention duration (β = -0.11, SE = 0.19, P = 0.564) had a statistically significant moderating effect on the improvement of depressive symptoms. These models exhibited very low explanatory power (adjusted R² ≈ 0%), suggesting that age and duration, as single variables, cannot adequately account for the observed heterogeneity. The non-significant findings might be attributed to the relatively limited number of studies included, which constrained the statistical power of the meta-regression, or to potential complex interactions with other factors (e.g., exercise type, intensity) (for detailed results, see Figures 7, 8).

In summary, these exploratory subgroup analyses suggest that moderate-intensity and mind-body exercises might be associated with improvements in depressive and anxiety symptoms. However, the meta-regression analyses did not detect significant linear moderating effects for age or intervention duration as continuous variables. Given the limited number of studies within certain subgroups and the substantial heterogeneity observed, these findings should be interpreted with caution and viewed as hypothesis-generating rather than definitive. The heterogeneity among studies may stem from other unmeasured factors, complex interactions among multiple variables, or the inherent limitations of the current sample size. These preliminary insights may inform the design of future research, but more robust evidence is needed before definitive clinical prescriptions can be made.

To assess the potential for publication bias in this meta-analysis, we performed both a visual inspection of the funnel plot asymmetry (Peters et al., 2008) and Egger’s linear regression test (Irwig et al., 1998) to evaluate publication bias statistically (Afonso et al., 2024).

The results indicated no significant evidence of publication bias. For depression, Egger’s test yielded a z-value of 0.39 (P = 0.742 > 0.05), suggesting that the funnel plot was largely symmetrical. For emotion regulation, the test resulted in a z-value of 0.49 (P = 0.504 > 0.05), also indicating good symmetry of the funnel plot. Similarly, for anxiety, the test produced a z-value of 1.05 (P = 0.230 > 0.05), demonstrating no significant asymmetry. These findings imply that the distribution of effect sizes from the included studies was relatively even for depression, emotion regulation, and anxiety. There was no systematic bias indicative of the preferential publication of studies with statistically significant results or larger effect sizes. Consequently, these results enhance the credibility and external validity of the findings from this meta-analysis (for detailed results, see Figures 9–11).

The pooled results of this meta-analysis indicate that exercise interventions significantly improved depressive symptoms (SMD = 0.42, 95% CI: 0.13–0.71) and emotion regulation ability (SMD = 0.47, 95% CI: 0.06–0.92) in individuals with ADHD, and showed a positive trend toward alleviating anxiety symptoms (SMD = 0.43, 95% CI: 0.006–0.84). This finding provides high-level evidence supporting the establishment of exercise intervention as a valuable non-pharmacological adjunctive approach within the comprehensive management of ADHD. The mental health benefits derived from exercise may stem from multi-level physiological and psychological mechanisms. Physiologically, aerobic exercise promotes the release of Brain-Derived Neurotrophic Factor (BDNF) and enhances neuroplasticity in the prefrontal cortex and limbic system (e.g., hippocampus, amygdala), which collectively regulate emotional processing and stress responses (Kandola et al., 2019). Concurrently, exercise modulates autonomic nervous system function, increasing parasympathetic tone, thereby reducing physiological arousal levels and alleviating emotional dysregulation and anxiety commonly seen in ADHD (Chang et al., 2012). From a psycho-behavioral perspective, exercise provides a predictable environment, clear rules, and immediate feedback for individuals with ADHD, which helps enhance their self-efficacy and sense of control over emotions (Xie et al., 2021). Social interactions within group-based sports also offer valuable opportunities for learning emotional expression and regulation skills (Xie et al., 2021).

Subgroup analyses revealed significant moderating effects of exercise prescription parameters on psychological benefits, providing critical insights for developing precise intervention protocols. Firstly, regarding exercise type, mind-body exercises (e.g., yoga, tai chi) demonstrated significant medium-to-large effect sizes for both depression (SMD = 0.46, 95% CI: 0.15–0.78, P = 0.004) and anxiety (SMD = 0.80, 95% CI: 0.56–1.61, P = 0.002), outperforming purely physical exercises (Li et al., 2023). This advantage may be attributed to the unique “mind-body integration” characteristic of these activities: they involve physical exertion while emphasizing awareness of breath, focused attention, and acceptance of present-moment physical and mental states (Li et al., 2023). This mode of practice directly trains core components of attention and emotion regulation—response inhibition and metacognitive monitoring—thus more directly targeting the pathophysiological core of emotional dysregulation in ADHD (Barkley and Fischer, 2010). Secondly, concerning exercise intensity, moderate-intensity exercise yielded the most significant improvement in depressive symptoms (SMD = 0.54, 95% CI: 0.18–0.90, P = 0.003), whereas the effects of low-intensity and high-intensity exercises were non-significant. This finding aligns with the “inverted-U” dose-response hypothesis (McMorris and Hale, 2012), which posits that moderate physiological stress optimally activates neuroendocrine systems (e.g., promoting BDNF secretion, modulating cortisol levels). In contrast, intensities that are too low may fail to induce sufficient neural adaptation, while those too high might lead to excessive fatigue and negative emotional experiences, thereby attenuating the intervention effect (Heijnen et al., 2016).

Subgroup analyses indicated that the exercise intervention significantly improved depressive symptoms in adolescents (SMD = 0.35, 95% CI: 0.01–0.70, P = 0.046), whereas the effect for children, although showing a moderate trend, did not reach statistical significance. This discrepancy may reflect differences in neuroplasticity and psychosocial needs across developmental stages. Adolescence is a period of high incidence for depressive symptoms, and cognitive abilities at this stage are sufficiently developed to better comprehend and benefit from the psychological advantages of exercise (Kessler et al., 2007). For children, exercise interventions might need to place greater emphasis on fun and gamification; the sensitivity of emotional outcome measures in this group might be lower compared to behavioral measures, which could partly explain the non-significant result (Neudecker et al., 2019). The negative coefficient for age in the meta-regression (albeit non-significant) also suggests that for older individuals with ADHD, intervention strategies might need adjustment, such as incorporating cognitive challenges or psychoeducation, to maintain efficacy. Furthermore, the role of intervention duration appears complex. Single sessions demonstrated immediate anxiety-reducing effects, while sustained, long-term interventions seem crucial for consolidating psychological benefits (Ashor et al., 2015). Future research, through well-designed trials, is needed to clarify the optimal duration and frequency for different age groups.

The positive effects of exercise on depression, anxiety, and emotion regulation in ADHD are underpinned by synergistic physiological changes across multiple systems. Neurochemically, exercise upregulates monoaminergic neurotransmitters (e.g., serotonin, norepinephrine, dopamine) in key emotion-regulation regions such as the prefrontal cortex, hippocampus, and amygdala. This directly improves neural transmission linked to mood enhancement (Dey et al., 1992). Concurrently, exercise-induced increases in BDNF expression promote neuronal survival and synaptic plasticity, counteracting stress-induced atrophy and providing a structural basis for long-term antidepressant effects (Alizadeh and Dehghanizade, 2022; Coll-Padrós et al., 2019). At the neural network level, evidence suggests exercise normalizes hyperactivity and aberrant connectivity in the Default Mode Network (DMN) and Salience Network (SN), which are often dysfunctional in ADHD (Mehren et al., 2019). This optimization constitutes the neural foundation for effective emotion regulation. Finally, exercise modulates autonomic function by increasing vagal tone and heart rate variability, thereby improving physiological hyper-reactivity to emotional stress (Cortese et al., 2012; Liang et al., 2019). Collectively, these multi-level adaptations—spanning neurotransmitter balance, neurotrophic support, network integration, and autonomic regulation—constitute the biological mechanisms through which exercise alleviates comorbid symptoms and enhances emotion regulation in ADHD (Lin and Kuo, 2013; Ontiveros et al., 2025).

The findings of this study provide evidence-based support for integrating exercise interventions into the clinical management of ADHD; however, these recommendations must be framed with caution considering the variability in intervention protocols and study quality observed in the existing literature. Specifically, while moderate-intensity aerobic exercise and mind-body exercises show promise for alleviating depression and anxiety, the heterogeneity in intensity measurement and duration across studies suggests that individual responses may vary. Therefore, exercise prescriptions should be personalized and closely monitored, acknowledging the current lack of standardized “dose-response” guidelines (Lin, 2021).

Crucially, it is essential to emphasize that exercise should be considered a complementary (adjunctive) intervention rather than a stand-alone treatment for mental health outcomes in ADHD. Given the complexity of ADHD and its comorbidities, exercise is most effective when integrated into a multimodal treatment plan that includes pharmacotherapy, behavioral therapy, or psychoeducation as indicated. Clinicians should avoid overgeneralizing these findings and instead view exercise as a valuable, evidence-based component of a broader, holistic management strategy aimed at improving the overall quality of life for individuals with ADHD. Future clinical implementation should involve interdisciplinary collaboration to tailor interventions to the developmental stage and symptom profile of the patient, while maintaining realistic expectations regarding the magnitude of psychological benefits.

Although this meta-analysis provides integrated evidence supporting exercise interventions for improving mental health in individuals with ADHD, several limitations warrant careful consideration when interpreting the conclusions, and these limitations also chart a course for future research (Mayer et al., 2025).

Firstly, on methodological grounds, the overall number of included studies and their sample sizes remain relatively limited. This may have constrained the statistical power in subgroup analyses, particularly in the meta-regression, to detect the true effects of moderating variables. Some subgroups (e.g., the effect of different intervention durations on emotion regulation) could not be analyzed in depth due to an insufficient number of studies. Future research needs to include more high-quality, large-sample randomized controlled trials to enhance the robustness of the evidence (Mayer et al., 2025).

Secondly, significant heterogeneity exists in the parameters of the exercise interventions across the available literature. For instance, exercise intensity was often based on subjective descriptions rather than objective, quantifiable measures like heart rate monitoring, and intervention duration and frequency varied widely. This inconsistency may confound the precise interpretation of the optimal “dose-response” relationship. Therefore, future studies should employ more standardized exercise prescriptions and objective quantification tools to clarify the optimal protocols (regarding type, intensity, frequency, and duration) for specific mental health outcomes.

Thirdly, the robustness of the pooled effect sizes is influenced by residual heterogeneity and the risk of bias. Despite conducting subgroup analyses, substantial unexplained heterogeneity remained in several outcomes, indicating that other unmeasured factors—such as instructor expertise, baseline symptom severity, or specific environmental contexts—may significantly influence intervention effects. This residual variability suggests that the pooled estimates represent an average effect that may not apply uniformly to all settings. Additionally, most included studies were rated as having a “moderate” risk of bias, largely due to the inherent challenge of blinding participants and personnel in non-pharmacological exercise trials. The lack of blinding can introduce performance and detection biases, potentially inflating the observed effect sizes. Consequently, while the overall trends are promising, the magnitude of the benefits reported here should be interpreted with these methodological constraints in mind.