The study encompasses the following questions: (1) What exercise intervention programs are currently applied to improve and delay cognitive frailty in older adults, and what are their specific contents? (2) Are exercise intervention programs effective? (3) Does the literature propose potential biological mechanisms underlying the effects of exercise interventions on cognitive frailty?

The literature search was conducted using PubMed, Cochrane Library, Embase, Web of Science, and Scopus databases. All relevant articles published from the inception of each database up to October 9, 2025, were searched using a combination of Medical Subject Headings (MeSH) terms and free-text keywords. To ensure comprehensive retrieval, additional studies were identified through cross-referencing within. Key concepts in the search include “exercise,” “cognitive frailty,” “frail older adult,” and “cognitive dysfunction.” Detailed search terms, concepts, and strategies are provided in Supplementary Appendix 1.

Inclusion criteria: (1) Participants aged ≥60 years diagnosed with or defined as having cognitive frailty. (2) Study designs: randomized controlled trials, quantitative studies, or mixed-methods studies. (3) Interventions comprised exercise interventions for older adults with cognitive frailty or combined interventions integrating exercise.

Exclusion criteria: (1) Studies involving participants exhibiting only frailty or cognitive impairment, or those with dementia or psychiatric disorders. (2) Literature types such as reviews, animal studies, protocols, conference abstracts or posters, and studies with inaccessible full texts. (3) Literature lacking detailed descriptions of exercise intervention content or outcome measures, or intervention protocols including confounding factors such as medication or physical therapy.

EndNote X9 was used to manage and screen retrieved literature. After removing duplicates, two trained researchers independently conducted an initial screening based on inclusion and exclusion criteria by reviewing titles and abstracts. Subsequently, they independently reviewed the full text for further screening. Any discrepancies were resolved through discussion with a third researcher until a consensus was reached.

Data were extracted from the included studies, covering the following: author, publication year, country, study design, diagnostic criteria for cognitive frailty, sample size, types of exercise intervention, and outcome measures. To further detail the exercise interventions, we also extracted specific parameters of the exercise protocols, including intervention venue, duration, frequency, intensity, and cycle. Additionally, we extracted data on potential mechanisms underlying the effects of exercise interventions on cognitive frailty. The first author created the data extraction form, which was reviewed and supplemented by the second author. Any discrepancies between the authors were resolved through discussion and consultation with a third researcher to ensure accuracy and consistency in the data extraction process.

To ensure consistency in the concept of cognitive frailty, all included studies adopted the definition of cognitive frailty in older adults proposed by the International Association for Nutrition and Aging (IANA) and the International Association of Gerontology and Geriatrics (IAGG) (1). However, due to the absence of a unified assessment tool for cognitive frailty, diagnostic criteria may vary across studies. As this study is a scoping review aimed at mapping and summarizing the effects of exercise interventions on cognitive frailty—rather than quantifying or eliminating statistical heterogeneity—such variations in diagnostic criteria reflect the conceptual and methodological diversity within the field. To ensure transparency, details regarding study design, diagnostic criteria, and sample characteristics are summarized in the results section.

Two researchers independently applied the Mixed Method Appraisal Tool (MMAT, 2018 version) (16) to assess the quality of included studies. This involved answering two screening questions and five criteria per study type (qualitative studies, quantitative randomized controlled trials, quantitative non-randomized studies, quantitative descriptive studies, mixed-method studies). After completion, evaluators cross-checked results, resolving any disagreements through discussion with a third researcher.

The initial search yielded 4,431 records. After duplicate removal, 1,576 articles were excluded, leaving 2,855. After screening titles and abstracts, 2,715 articles were excluded, leaving 140 documents. Following a secondary screening, 30 articles with unavailable full texts were excluded, resulting in 110 remaining documents. After further full-text evaluation, 45 articles were excluded for non-compliance with the research theme, 14 for mismatched literature types, and 34 for inappropriate study populations. Ultimately, 17 studies met the inclusion criteria (Figure 1).

This scoping review included 17 studies published between 2018 and 2025 from four countries: China (n = 14), the United States (n = 1), South Korea (n = 1), and Canada (n = 1). The included studies comprised 16 randomized controlled trials and one quasi-experimental study. Tables 1, 2 presents the key details of the included studies.

The included studies were generally of high quality, with one study rated as moderate quality. See Supplementary Appendix 1 for details. The overall evidence demonstrated good consistency and robustness, which did not affect the systematic integration of exercise intervention effects and mechanisms in this study.

Currently, there are no unified diagnostic criteria for cognitive frailty. Researchers predominantly use a combination of physical frailty and cognitive function assessment scales to screen older adults for cognitive frailty. Among these, the most commonly employed combinations include: the Fried frailty phenotype (FP)combined with the Clinical Dementia Rating (CDR) scale (18, 24, 25), the FP combined with the Montreal Cognitive Assessment (MoCA) (19, 22, 30), the FP combined with both CDR and MoCA (17, 31, 32), and the Edmonton Frailty Scale (EFS) combined with MoCA and the Global Deterioration Scale (GDS) (26, 27, 29). However, even within the same combination, variations exist in the versions of the scales used and their respective scoring criteria.

We also extracted and summarized the mechanisms by which exercise interventions influence cognitive frailty as described in the included studies. Seven studies briefly explained potential mechanisms, primarily encompassing five types. (1) Brain Structure and Plasticity: Exercise interventions delay volume atrophy in hippocampal subregions (left parasubiculum, left HATA, right CA1, and right presubiculum) (26), enhance functional connectivity within the prefrontal-hippocampal network, and improve cognitive function by modulating brain structural and functional plasticity. This effectively slows memory decline in older adults.(2) Prefrontal Activation and Neural Efficiency: Exercise intervention enhances prefrontal cortical activation levels (33), influencing neural efficiency to promote shorter reaction times and improved accuracy. (3) Neuroinflammation and Oxidative Stress: Exercise intervention regulates inflammatory cytokine levels (IFN-γ, IL-2, IL-4), suppresses inflammasome activation (28), enhances antioxidant enzyme activity (20), and reduces oxidative stress-induced neuronal damage. (4) Neurotransmitter changes: Exercise intervention enhances neuronal plasticity by upregulating brain-derived neurotrophic factor levels (21). (5) Improved Cerebral Blood Flow and Metabolism: Exercise increases blood flow parameters in key cerebral vessels, including peak systolic velocity (PSV), mean blood flow velocity (MBFV), and end-diastolic velocity (EDV) in the right middle cerebral artery, PSV in the left middle cerebral artery, and MBFV and EDV in the basilar artery (27), thereby enhancing cerebral oxygen supply. It also reduces glycated hemoglobin levels (30), contributing to delayed progression of cognitive frailty (Table 3).

Regarding exercise intervention formats, they are primarily categorized into offline and blended online-offline approaches. Face-to-face offline interventions demonstrate significant advantages in real-time guidance and structured supervision, aiding in the prevention of fall risks (23). Furthermore, offline interventions maintain robust social interaction support, enhancing psychological belonging (22). However, limitations in accessibility and long-term monitoring may arise due to geographical and temporal constraints. Against this backdrop, digital technology—leveraging powerful exercise parameter analysis, real-time remote guidance capabilities, and immersive application scenarios—has deeply integrated into exercise interventions, offering a more effective and sustainable practice paradigm for improving cognitive frailty in older adults. Although digital exercise offers unique advantages in quantifying movement data, enhancing training motivation, and overcoming temporal and spatial constraints, barriers remain in widespread adoption and long-term efficacy validation. This stems from the fact that digital exercise programs require participants to engage in sustained self-management at home. In this context, the complexity of digital technology usage may pose learning and acceptance challenges for older adults (34), potentially exacerbating their feelings of loneliness (24).

Therefore, further efforts should be made to integrate user-friendly digital interfaces, simplify functional designs to meet the needs of diverse populations, and strengthen technology training for older adults, thereby optimizing the acceptability and adaptability of digital technology. Additionally, future efforts should fully integrate online and offline strengths: establishing trust foundations through offline components, then leveraging digital tools for long-term tracking and dynamic adjustments to enhance intervention accessibility and continuity. More importantly, real-time, multidimensional biofeedback systems must be strengthened. By leveraging an integrated framework of “exercise formats-mechanisms-health outcomes” dynamic models linking exercise parameters, mechanism pathways, and health outcomes should be developed. This will propel digital exercise toward a truly personalized, mechanism-driven intervention paradigm.

The effectiveness of exercise interventions in improving cognitive frailty stems from their underlying biological mechanisms involving multi-pathway, multi-level synergistic effects. Aerobic exercise has been demonstrated to effectively enhance peak oxygen uptake, strengthen cardiopulmonary function, and consequently significantly increase cerebral blood flow while reducing central arterial stiffness (35), thereby promoting cognitive function improvement. Resistance training activates the bone-brain axis (36), stimulating skeletal muscle to release and upregulate irisin. This process elevates levels of IGF-1 and BDNF, thereby delaying skeletal muscle atrophy and promoting neuroplasticity and cognitive enhancement. Traditional fitness qigong exercises, as mind–body exercises integrating intention, breath, and movement, not only mitigate hippocampal subregion atrophy (26, 37) and but also induce neurogenesis and synaptic plasticity. Furthermore, it exerts neuroprotective effects by modulating oxidative stress and inflammation through reducing pro-oxidative markers (MDA, 8-iso-PGF2α), enhancing antioxidant enzyme SOD activity, and regulating inflammatory cytokine expression (IFN-γ, IL-2, IL-4) (28). Dual-task training simultaneously influences peripheral metabolism and central nervous system efficiency: peripherally, it reduces HbA1c levels (30) and alleviates metabolic stress; centrally, it enhances neural information processing efficiency and improves cognitive function by decreasing oxygenated hemoglobin concentration and activation levels in the prefrontal cortex (33). Multicomponent exercise, integrating aerobic exercise, resistance training, and balance training, further generates synergistic benefits across neural (21), metabolic (22), and immune (38) systems.

Collectively, these mechanisms illustrate that exercise interventions improve cognitive frailty not through a single pathway, but via the coordinated regulation of multiple, interconnected biological systems. These pathways span various dimensions, including neural plasticity, modulation of inflammation and oxidative stress, optimization of cerebral blood flow and metabolism, and neurotrophic support, forming a systemic adaptive network characterized by “peripheral–central–behavioral” integration. However, the literature included in this review offers limited mechanistic exploration. Only a portion of studies employed biomarker assessments, and objective neuroimaging evidence remains generally lacking. Future research must strengthen neuroimaging and biomarker analyses to identify potential biological targets for personalized interventions.

The studies included in this review indicate that exercise interventions for cognitively frail older adults encompass diverse approaches. Due to variations in intervention content and delivery, the effectiveness of exercise interventions in improving cognitive frailty among older adults also differs. Even among studies employing identical exercise programs, this review observed divergent intervention outcomes. Yoon et al. (18) demonstrated that 16 weeks of moderate-intensity high-speed resistance training significantly improved participants’ processing speed and executive function. Conversely, Wu et al. (19) found no significant effect on executive function, potentially due to the low intensity and short duration of their intervention, which may have been insufficient to activate the PI3K-AKT–mTOR pathway, thereby limiting adaptive remodeling in muscle and neural cells (39). In contrast, dual-task training (30–33)—by incorporating detailed motor parameters and simultaneously engaging peripheral and central systems—not only enhances coordination between physical activity and cognitive tasks but also accelerates cognitive switching speed, thereby boosting the brain’s adaptability and flexibility. Traditional fitness qigong exercises (25–30), as the primary exercise type for improving cognitive frailty, demonstrated more significant simultaneous improvements in both physical and mental health compared to single-modality aerobic exercise (17) and resistance training (18, 19). This may be attributed to its emphasis on the holistic coordination mechanism of “regulating the body, regulating the breath, and regulating the mind,” which integrates functional restoration with psychological regulation. Consequently, it demonstrates unique intervention advantages across multidimensional health outcomes.

Although these findings align with existing research (40), supporting the positive association between exercise dosage, intensity, and pattern diversity with cognitive function. However, our review reveals significant variability in the design of current exercise intervention parameters—including intensity, frequency, and duration—lacking systematicity and comprehensiveness. This limitation constrains the potential for greater improvement in cognitive frailty. Therefore, future exercise intervention designs should thoroughly explore optimal training ratios and progression models within consistent intervention frameworks. This approach will enable the optimization of exercise content and load, thereby promoting comprehensive improvements in patient health outcomes.

Despite highlighting the positive role of exercise interventions in improving cognitive frailty, this review also acknowledges several limitations in the current evidence. First, most of the included studies were conducted in China, with limited representation from other regions, and generally featured small sample sizes, which may constrain the generalizability of the results. Therefore, future research should employ cross-cultural, multicenter, large-scale randomized controlled trials to validate the feasibility and applicability of different intervention protocols.

Second, the studies included in the review employed diverse diagnostic criteria and measurement tools. Although all reported significant improvements in cognitive frailty following exercise interventions, there was no uniformity in the target populations or settings. Furthermore, insufficient research exists on mental health outcomes, thereby undermining the standardization and comparability of findings. Future efforts should establish standardized diagnostic and assessment systems. This should involve integrating objective criteria to unify evaluation tools and applicable standards across different research settings, while strengthening the measurement of multifaceted health outcomes. Such measures will ensure the comparability, applicability, and practical utility of research findings.

Third, existing studies predominantly employ screening scales as primary outcome measures, with only a few incorporating biomarkers for corroboration, yet evidence remains limited. Future efforts should focus on developing a multidimensional, integrated assessment framework that organically combines functional screening, biomarkers, and multimodal neuroimaging to comprehensively elucidate intervention effects and underlying mechanisms.

Moreover, the integration of digital technology into exercise interventions remains relatively underdeveloped. Moving forward, we should fully leverage the potential of digital tools, relying on multidimensional bioinformation integration and real-time feedback technologies to continuously analyze and model participants’ exercise data. This will enable dynamic optimization of exercise programs, delivering precision and personalization in interventions.