Evaluating the cognitive impact of exergames on community-dwelling older adults beyond laboratory settings: a systematic review and meta-analysis

1 Introduction

Aging-related cognitive decline has raised significant concerns worldwide, and this decline typically affects areas such as memory, attention, processing speed, and executive function. According to the WHO's 2023 dementia report, over 55 million people are living with dementia, with nearly 10 million new cases emerging each year (World Health Organization, 2023). As the global population ages, this increase contributes to the global burden of disease (Nichols et al., 2022), affects individuals' quality of life, and heightens the stress experienced by informal caregivers within families (Chiao et al., 2015), and substantial economic costs for healthcare systems and societies (Nichols et al., 2022; Wimo et al., 2017).

Current pharmaceutical treatments for cognitive impairment focus primarily on symptom relief without modifying the underlying disease (Srivastava et al., 2021). Researchers have explored various non-pharmaceutical interventions, such as lifestyle interventions, aimed at preventing or slowing cognitive decline and enhancing cognitive performance in older adults. Several lifestyle interventions, such as social engagement, nutrition, and sleep hygiene, have also been implicated in maintaining cognitive health (Kivipelto et al., 2018). These lifestyle interventions have the advantage of fewer side effects and can be used alongside pharmaceutical treatments without adverse interactions (Sikkes et al., 2021).

Aerobic exercise (AE) and cognitive training (CT), in particular, have shown robust benefits for cognitive function across the lifespan (Erickson et al., 2019; Smith et al., 2009). While physical exercise and cognitive training have often been studied separately as single-domain approaches, numerous studies have demonstrated that the combination of exercise and cognitive training is more effective in enhancing cognitive function than either exercise or cognitive training alone (Schaefer and Schumacher, 2010; O'Dwyer et al., 2007; Law et al., 2014; Wollesen and Voelcker-Rehage, 2014). In recent years, there has been a growing trend of using games to combine AE and CT training (Law et al., 2014; Ghai et al., 2017; Mancioppi et al., 2021; Qian et al., 2024). Game-based training programs provide interactive systems (Lumsden et al., 2016) and timely feedback (Kappen et al., 2016), making training more accessible and available to use for older adults in both clinical and home settings (Hill et al., 2017; Gavelin et al., 2021; Zhao et al., 2020).

Exercise games, which are interactive games that require physical movements to play, represent a popular and engaging form of combined physical-cognitive training (Mueller et al., 2018). These games often incorporate elements of balance, coordination, and cognitive challenges within a virtual environment. Exercise games have been increasingly utilized as a physical-cognitive tool due to their interactive nature, which can seamlessly integrate physical and cognitive training together in an entertaining way. Their effectiveness is supported by a series of combined physical and cognitive training studies (Law et al., 2014; Ghai et al., 2017; Mancioppi et al., 2021). Commercial gaming platforms like the Nintendo Wii and Xbox Kinect have been shown to demand physical effort comparable to traditional moderate exercise (Miyachi et al., 2010; Lyons et al., 2011; O'Donovan et al., 2012; Taylor et al., 2012). Moreover, recent virtual reality exercise games, such as the Supernatural boxing-based workout, have demonstrated even higher energy expenditure than traditional game platforms (Gomez et al., 2018; Craig et al., 2024). Beyond the physical benefits of exergaming, which is generally defined as a digital game in which the outcome of the game is predominantly determined by physical effort (Mueller et al., 2011), older adults also experience cognitive gains from the game (Anguera et al., 2022). Various systems have been used for exercise game interventions, and studies have consistently shown that these systems offer equal or superior cognitive benefits compared to single-domain and combined physical-cognitive training (Howes et al., 2017; Stanmore et al., 2017).

Although early evidence on exercise games' cognitive benefits is promising, it is mainly derived from lab or clinical settings. To address the growing demand for community-based interventions and build on the existing evidence base, we conducted a systematic review and meta-analysis to examine the available evidence from community-based studies beyond lab settings. In this meta-analysis, we investigate the overall impact of exercise games on cognition for community-dwelling older adults, which includes both non-clinical and clinical populations (e.g., mild cognitive impairment and dementia). By examining the influence of exercise game characteristics, clinical- and non-clinical population, this study seeks to provide valuable insights for designers, developers, and researchers.

2 Method

We adhered to the Cochrane Handbook for Systematic Reviews of Interventions for data processing and analysis in the meta-analysis. Additionally, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; www.prisma-statement.org) guidelines for reporting our findings. The review is registered on the Open Science Framework (OSF) https://doi.org/10.17605/OSF.IO/AM8QT.

2.1 Search strategy

A systematic electronic database search was conducted to identify relevant published studies. The PubMed, ScienceDirect, SpringerLink, ACM, IEEE databases were included in our search to identify relevant studies published from Jan, 2014 to June, 2024.

For this review of the impact of exercise games on cognitive function in older adults, a comprehensive literature search was conducted using specific keywords and Boolean operators to ensure a detailed and focused retrieval of relevant studies. The search terms included combinations of (“exercise game" OR “active video game" OR “virtual reality" OR “augmented reality") to capture various forms of interactive and immersive exercise technologies. These were paired with cognitive function-related terms (“attention" OR “memory" OR “executive function" OR “processing speed") to target studies examining specific cognitive outcomes. Additionally, the search was refined to the population of interest using terms like (“older adults" OR “seniors" OR “elderly"). The detailed search terms for each database are described in Supplementary material Data Sheet 1.

2.2 Study selection

Our meta-analysis only included studies published in peer-reviewed journals or at conferences. Posters presented at peer-reviewed conferences with an accessible full conference paper were also included. Protocols and literature reviews were excluded. Study participants included cognitively healthy older adults or older adults with cognitive impairments, such as Mild Cognitive Impairment (MCI) and different types of Dementia, and the mean age of participants had to be at least 60 years old. The studies included at least one exercise game training group, and studies were considered only when they included an active (e.g., yoga, running, biking) or non-active control group (e.g., educational material) in addition to the exercise game interventions group. Studies with interventions that were irrelevant to exercise game interventions were excluded from this meta-analysis. Studies needed to include at least 12 weeks of intervention. This threshold was chosen based on its widespread use in interventional studies focused on cognitive improvement in older adults (Ngandu et al., 2015; Dedeyne et al., 2017). Studies have demonstrated that cognitive changes are typically observable with 12 or more weeks to produce reliable and sustained measurements on cognitive improvements (Dedeyne et al., 2017). Studies with the acute effects of an intervention (e.g., after single training sessions) were excluded. Included studies could be randomized controlled trials (RCT) or cluster-RCT, and training had to take place in home-based or community-based environments (e.g., daycare center, nursing home, and other residential places), while lab-based or hospital- or clinic-based studies were excluded. Finally, all included studies must have the necessary data to compute Hedges' g, such as sample size, mean, standard deviation, and standard errors for at least one cognitive outcome, and the study had to include at least one cognitive measurement, which could be global cognition, attention, memory, executive functions, verbal learning, or processing speed.

2.3 Data extraction

Articles searched from electronic databases were organized into an Excel sheet with titles and abstracts for initial screening by two researchers. Two researchers (YQ and AS) extracted data independently from eligible studies and checked with each other, any disagreements were discussed with the third researcher (LC). We extracted the sample size, intervention, and control group characteristics, as well as the necessary raw data that were computed for the effect size (Hedges' g), such as mean, median, standard deviation, and standard error of pre- and post-cognitive outcomes. For median, range, and sample size, we used Hozo's method to transform the median and range to mean and standard deviation (Hozo et al., 2005). If the study was relevant to our analysis but the data necessary to calculate the effect sizes were missing, the authors were contacted via email to obtain the relevant data.

2.4 Outcome measures

Based on published reviews and factor analyses (Miyake et al., 2000; Friedman and Miyake, 2004; Rieker et al., 2022), the cognitive outcomes included objectively assessed cognitive domains of global cognition, executive control, memory, attention, processing speed, and verbal abilities. The results of cognitive screening tools, such as MoCA and MMSE, are classified as global cognition. The classification of executive functions included tests that measured working memory, inhibition, and cognitive flexibility. Memory included short- and long-term memory measurements. Attention included divided, selective, and sustained attention measures. Processing speed included tests that measured reaction times. Verbal fluency included assessments of verbal, categorical, and phonological fluency.

2.5 Risk of bias

Two authors (YQ and AS) evaluated the risk of bias independently in qualified studies using the Cochrane risk of bias in randomized trials tool (ROB-2), which includes five domains (1) Risk of bias arising from the randomization process; (2) Risk of bias due to deviations from the intended interventions; (3) Missing outcome data; (4) Risk of bias in measurement of the outcome; (5) Risk of bias in selection of the reported result. Disagreement was resolved through consultation between authors.

2.6 Statistical analyses

Raw data (sample size, mean, and standard deviation) of post-intervention outcomes across studies were collected and organized in an Excel spreadsheet, exported as a CSV format file, and then imported into R. All outcome measurements were treated as continuous variables. The meta-analysis was conducted in R Statistical Software (v4.4.1) with R packages meta (Balduzzi et al., 2019), metafor (Viechtbauer, 2010), dmetar (Harrer et al., 2019), clubSandwich (Pustejovsky, 2024), and esc (Lüdecke, 2019). A random-effects model was used due to anticipated variability between studies and corrected standardized mean difference (Hedges' g) and 95% confidence intervals (CI) were calculated to compare post-intervention outcomes between the exergaming and control groups. When multiple outcomes within the same cognitive domain were reported by a single study, effect sizes were aggregated at the study level using inverse-variance weighting to maintain statistical independence.

Heterogeneity was assessed using the I² statistic and Cochran's Q test (χ² p-value). The I² statistic quantifies the percentage of variability in effect sizes attributable to true between-study heterogeneity rather than sampling error, with values of 25%, 50%, and 75% indicating low, moderate, and substantial heterogeneity, respectively (Higgins et al., 2003). A Cochran's Q p-value of less than 0.05 indicates significant heterogeneity. Restricted maximum likelihood estimation (REML) was used to estimate the between-study variance τ² (Viechtbauer, 2005). Knapp-Hartung adjustments (Knapp and Hartung, 2003) were applied to calculate confidence intervals around pooled effects, which provides more conservative estimates particularly when the number of studies is small. The resulting t-statistic and associated p-value test the null hypothesis that the true pooled effect equals zero. For cognitive domains with only one study (k = 1), heterogeneity statistics could not be calculated, and the reported effect represents a single study finding rather than a meta-analytic estimate.

Between-study heterogeneity can be caused by one or more studies with extreme effect sizes, which may distort the pooled effect estimate of the meta-analysis (Harrer et al., 2021). Thus, outlier and influential case analyses were performed to identify studies or data points that have a strong influence on overall effect sizes. The function find.outlier in R package dmetar was applied to an object input created by an R meta package meta-analysis function. Baujat charts, the leave-one-out method, and funnel plots were used for outlier and influential case analysis; the leave-one-out method combined with forest plots provided recalculated pooled effects with one study omitted each time. Furthermore, funnel plots were used to visualize the symmetrical pattern of publications, with Egger's regression test (Egger et al., 1997) conducted to assess potential publication bias.

2.7 Subgroups analyses

Subgroup analyses were performed separately for cognitive domains (attention, executive function, global cognition, memory, verbal ability, and processing speed), population types (cognitively healthy older adults, people with dementia, and mild cognitive impairment), and exercise modes (aerobic, non-aerobic, and mixed), with appropriate aggregation applied to maintain one effect size per study within each subgroup category. All statistical tests used a significance level of p < 0.05.

3 Results

3.1 Database search

A total of 7,257 articles were found in the databases after we applied search terms (Figure 1). After removing duplications and non-intervention studies, such as reviews, workshops, case studies, and protocols, 6143 articles were left for screening. Of these, 5,666 articles were irrelevant to our interests based on the initial review of the title and abstract, and the irrelevant research include the ones focused on application, device development, or studies in different fields (e.g., K12 education, economics or sports, etc). Full text screening was performed with the remaining 477 studies, we screened out further studies due to irrelevant study design (mean age < 60, non-community-based environment, no randomized controlled trial, and not multiple session training). Three additional studies were excluded from the meta-analysis due to the need to report necessary data to compute the effect sizes. In total, our selection criteria left 10 studies for this meta-analysis (Stanmore et al., 2019; Martel et al., 2018; Gouveia et al., 2021; Gschwind et al., 2015b,a; Karssemeijer et al., 2019; Wu et al., 2023; Liao et al., 2021, 2020; Van Santen et al., 2020).

A total of 690 (EXG = 350, Control = 340) older adults participated in these 10 studies (Table 1). The mean age of the exercise game intervention group is 74.11 (SD = 21.29), and the mean age of the control group is 74.71 (SD = 21.19). One study does not report the exact mean age for each group but only a general range (e.g., 61–66, 71–75) so this study is excluded from calculating the overall mean age. Among all participants, 80 participants have mild cognitive impairment (MCI), 192 participants are people with dementia (PwD), and 418 participants are cognitively healthy older adults. Of all studies, only three studies include a follow-up assessment. Thus, the post-follow-up analysis is not included in this meta-analysis.

Table 1. Summary of studies.

EI, Exergaming group; CT, Control group; Pop., Population (Hlth., Healthy; PwD, People with Dementia); Type = exercise game Type (Aero., Aerobic; Comb., Combined; Non-aero., Non-aerobic); RT, Reaction Time; LT, Long-term; ST, Short-term. TMT-A/TMT-B, Trial Making Test; ANT, Attention Network Test; MoCA, Montreal Cognitive Assessment; MMSE, Mini-Mental State Examination; ACE III, Addenbrooke's Cognitive Examination; CCVLT, Chinese Version Verbal Learning Test; COGTEL, Cognitive Telephone Screening Instrument; EXIT-25, Executive Interview.

The intervention training ranges from 12 to 24 weeks; seven studies conducted a 12-week intervention, two studies did a 16-week intervention, and one did a 24-week intervention. In terms of the frequency of the sessions, all studies chose to do either two sessions per week or three sessions per week. The duration of each session varies substantially among the studies, ranging from 20 to 150 minutes per session.

3.2 Risk of bias

Among the 10 studies (Table 2), six are categorized as “low risk,” two studies are at “some concerns,” and two studies are at “high risk.” The weighted bias figure is shown below (Figure 2). For all five bias domains, the overall bias level is low in our meta-analysis.

Table 2. The risk of bias evaluation on each study by risk domains.

D1 - Risk of bias arising from the randomization process; D2 - Risk of bias due to deviations from the intended interventions; D3 - Missing outcome data; D4 - Risk of bias in the measurement of the outcome; D5 - Risk of bias in the selection of the reported result.

3.3 Exergame design and implementation characteristics

The analysis of exergame features across the included studies (n = 10) revealed substantial heterogeneity in the completeness and depth of game design reporting (see Supplementary Tables S1, S2). Overall, the description of exergame characteristics was minimal across most studies, with reporting predominantly focused on movement mechanics and input-output mapping rather than comprehensive game design elements.

The majority of studies (8/10, 80%) primarily described exergames through their physical movement requirements and corresponding on-screen actions. Critical game design components were rarely reported across the reviewed studies. No studies explicitly described graphical style elements such as 2D versus 3D rendering, realistic versus stylized art direction, color palettes, or visual themes. Core game mechanics including scoring systems, difficulty progression algorithms, level structures, win or loss conditions, or reward mechanisms were largely unreported. Elements known to enhance player engagement—including competition modes (individual vs. multiplayer), progress tracking systems, achievement systems, narrative elements, or personalization options—were absent from most descriptions. While some studies mentioned “visual and auditory feedback” (Liao et al., 2020, 2021; Martel et al., 2018), the specific nature, timing, or purpose of this feedback remained unspecified. Only two studies provided slightly more comprehensive descriptions (Gschwind et al., 2015a; Gouveia et al., 2021)

3.4 Feasbility and safety

Attrition rates across the included studies ranged from 10.5% to 50% (median = 17.65%), see Supplementary Tables S1, S2, with most withdrawals attributed to personal health issues unrelated to the intervention, changes in living arrangements, or death, while a minority of cases cited low motivation or technical difficulties as reasons for discontinuation. Notably, studies explicitly reporting motivation-related attrition included (Liao et al., 2021), where two participants found the program “not attractive to them,” (Van Santen et al., 2020), where seven participants wanted to stop, and (Gschwind et al. 2015a), which reported system failures and technical issues as barriers. All included studies reported no adverse events related to the exergaming interventions, with (Gschwind et al. 2015a) and (Gouveia et al. 2021) specifically noting no falls or injuries associated with gameplay. These findings suggest that home-based and community-based exergaming interventions are both feasible and safe for older adults; however, enhancing game playability to increase enjoyment and simplifying technical setup to improve accessibility remain critical considerations for sustaining participant engagement and minimizing intervention-related attrition.

3.5 Effect of exercise game on cognition

Exercise game interventions produced a small but significant overall effect on cognitive function (g = 0.135, 95% CI [0.037, 0.229], p = 0.012, I² = 37.7%) (Table 3, Figure 3). Domain-specific analyses revealed non-significant effects for global cognition (g = 0.185, p = 0.240), executive function (g = 0.054, p = 0.473), memory (g = 0.179, p = 0.060), and attention (g = 0.124, p = 0.118), though memory approached statistical significance (Figure 4a). Processing speed showed a significant effect based on one study (g = 0.337, p = 0.018), while verbal function results were inconclusive due to high heterogeneity (I² = 72.6%) and imprecise estimates (Figure 4b).

Table 3. Effect of exercise game in comparison to control group on cognition.

MCI, People with mild cognitive impairment; PwD, People with dementia.

^*indicates p < 0.05;

^**indicates p < 0.01.

3.6 Subgroup: exercise game types

Aerobic exergaming yielded a small positive effect that approached but did not reach statistical significance (g = 0.119, p = 0.142), with moderate heterogeneity (I² = 40.5%, Q = 6.72, p = 0.152). Combined exercise programs demonstrated a small-to-moderate effects (g = 0.291, p = 0.138), with moderate heterogeneity (I² = 47.1%, Q = 3.78, p = 0.151). Non-aerobic exergaming showed a negligible but significant effect (g = 0.091, p = 0.007^**), without evidence of heterogeneity (I² = 0%, Q = 0, p = 0.985). The test for subgroup difference was not statistically significant (Q = 2.91, p = 0.233), suggesting that intervention effects did not vary significantly by exercise modality (Figure 5).

3.7 Subgroup: population types

Subgroup analyses by population type included 10 studies across three categories. Cognitively healthy older adults showed a small positive effect that did not reach statistical significance (g = 0.149, p = 0.101), with moderate but not significant heterogeneity (I² = 39.4%, Q = 6.60, p = 0.158). Individuals with mild cognitive impairment (MCI) demonstrated a small-to-moderate effect that was not statistically significant (g = 0.256, p = 0.168), with no evidence of heterogeneity (I² = 0%, Q = 0.59, p = 0.443). People with dementia (PwD) (k = 3) showed virtually no effect (g = 0.005, p = 0.975), with substantial heterogeneity (I² = 59.5%, Q = 4.60, p = 0.101). The test for subgroup differences was not statistically significant (Q = 3.21, p = 0.201) (Figure 6).

4 Discussion

This meta-analysis investigated the effects of exercise game interventions on the cognitive functions of older adults who live in communities with or without assisted facilities. It includes 10 eligible studies that explored the differential effects of exercise game interventions versus control groups. The included studies varied in study design, game platforms, and the specific exercise games employed. The control groups also differed, encompassing active, non-active, and traditional exercise interventions.

4.1 The effects of exercise games on cognition

In contrast to traditional cognitive training, which often exhibits limited transfer effects (Owen et al., 2010; Peckham, 2021), our findings demonstrate that exercise games exert a broader influence on cognitive function among community-dwelling older adults. Exercise games demonstrated preliminary evidence of improvements on memory and processing speed, and these results align with previous research highlighting the positive impact of exercise games on short-term memory (Roig et al., 2013) and processing speed (Moret et al., 2022). It could be that the inherent cognitive demands of exercise games may stimulate neural pathways in a manner that is distinct from traditional physical exercise (Eggenberger et al., 2016; Liao et al., 2019), potentially leading to unique neurophysiological adaptations and enhanced cognitive outcomes. However, global cognition, executive function, attention, and verbal fluency did not display significant improvements. It is worth mentioning that the relatively small number of studies focusing on processing speed and verbal fluency precluded definitive conclusions regarding the impact of exercise games on these domains.

In addition, the available outcome assessments only tap a small part of the broader constructs of cognitive functions, such as attention and verbal fluency. This led to another issue, for which accurate assessment of cognition remains a critical challenge in a community-based environment. Also, the inconsistent application of cognitive assessment tools, as exemplified by the varied categorization of the TMT-A task, can confound interpretation. Moreover, the use of screening tools such as the MoCA and MMSE as outcome assessments creates limitations for our interpretation of the findings, especially for the study that used screening tools as the only cognitive measurement. This point underscores the need for more comprehensive cognitive evaluations with detailed cognitive profiles to better characterize the outcomes. Identifying readily accessible yet reliable cognitive assessment tools for community-based research remains an ongoing challenge.

4.2 The effects of exercise games on non-clinical and clinical populations

Our meta-analysis examined the cognitive effects of exercise games across cognitively healthy older adults, individuals with mild cognitive impairment (MCI), and those with dementia (PwD). However, the evidence base for clinical populations remains severely limited and does not support definitive conclusions. With only two studies examining MCI participants and three studies examining dementia, none of these clinical subgroups demonstrated statistically significant effects (MCI: g = 0.256, p = 0.168; PwD: g = 0.005, p = 0.975). While the MCI subgroup showed a numerically larger effect size compared to healthy controls, this difference was not statistically significant and is based on an insufficient number of studies to draw meaningful inferences. Given the substantial heterogeneity in dementia studies (I² = 59.5%) and the non-significant test for subgroup differences (Q = 3.21, p = 0.201), we cannot determine whether exercise games produce differential effects across population types. Previous hypotheses, such as the “compensation hypothesis” suggesting greater neuroplasticity potential in individuals with lower baseline cognitive function (Sanders et al., 2019; Wu et al., 2019), or neurobiological constraints in advanced cognitive decline (Arnold et al., 2013; Buschert et al., 2011), remain speculative in the context of exergaming interventions due to insufficient empirical evidence. The critical gap in the current literature is the lack of adequately powered randomized controlled trials specifically enrolling individuals with MCI and dementia (Voinescu et al., 2021). Future research must prioritize well-designed studies with sufficient sample sizes in these clinical populations, employ comprehensive and standardized cognitive assessment batteries, systematically document intervention fidelity and adherence, and report feasibility and safety outcomes to establish whether exergaming interventions offer meaningful cognitive benefits for individuals with cognitive impairment in community settings.

4.3 The effects of exercise games by exercise type

Our subgroup analysis revealed that the cognitive benefits derived from exercise game interventions vary according to exercise type. Exercise games that integrated aerobic, resistance, and balance components yielded the most pronounced cognitive benefits. Aerobic exercise game intervention alone shows a small positive impact. In contrast, non-aerobic exercise alone did not produce substantial cognitive improvements.

Extensive research supports the positive influence of aerobic exercise and resistance training on cognition (Colcombe et al., 2006; Hillman et al., 2008; Erickson et al., 2009; Huang et al., 2022; Li et al., 2018). However, the impact of non-aerobic training, such as resistance exercises, on cognitive function is less well-established. Some studies have reported benefits for specific cognitive domains, but the evidence is inconsistent (Rogge et al., 2017; Dorner et al., 2007). Our findings align with a recent review suggesting that multicomponent exercise regimens, incorporating aerobic elements, are essential for optimizing cognitive outcomes in older adults (Yan et al., 2023). While the optimal exercise modality for enhancing cognitive function in older adults remains a subject of ongoing debate, our results underscore the importance of integrating aerobic components into exercise game interventions to maximize cognitive benefits. There is a new direction for developing games that blend cognitive and physical activity to promote the cognitive health of older adults as some current studies show (Anguera et al., 2022; Qian et al., 2024).

4.4 Implications for stakeholders

Exercise games represent a promising intervention modality for mitigating age-related physical and cognitive decline in older adults. A key advantage of exercise games lies in their scalability and accessibility, making them uniquely well-suited for home- and community-based interventions. However, our findings reveal a notable discrepancy: while laboratory-based studies have demonstrated significant moderate to large effect sizes for exercise game interventions (Yen and Chiu, 2021; Stanmore et al., 2017; Peng et al., 2024), community-based implementations show considerably attenuated clinical effects.

Understanding the factors contributing to this reduced efficacy in community settings is critical for optimizing the translation of exercise game interventions from controlled research environments to real-world applications. Several hypotheses warrant consideration. First, the mode of delivery and implementation fidelity may differ substantially between laboratory and community contexts, potentially affecting adherence, intensity, and overall intervention quality. Second, the characteristics of the exercise games themselves may play a pivotal role in determining outcomes. However, the majority of studies in this review focus primarily on reporting physical movement requirements while neglecting core game design elements—such as progress tracking, timing mechanics, competition, and scoring systems—that are integral to sustaining cognitive engagement. This reporting gap limits our understanding of why certain games may be more effective for specific cognitive domains than others. Moreover, without sufficient access to detailed game design specifications, important questions remain unresolved: Do specific cognitive elements embedded within games differentially affect particular cognitive functions? To what extent are observed cognitive benefits attributable to the physical exercise component versus the cognitive demands inherent in gameplay? Addressing these questions is essential for disentangling the mechanisms of action underlying exergame efficacy and for informing the evidence-based design of future interventions.

Notably, our meta-analysis reveals an important distinction from previous reviews examining commercial exergaming platforms (e.g., Nintendo Wii, VR Dance, and Xbox Kinect Sports) (Yen and Chiu, 2021; Stanmore et al., 2017; Peng et al., 2024) for older adults. The majority of studies included in the present analysis employed researcher-developed games rather than commercially available products. This raises a critical question: do the underlying game mechanics and design principles influence cognitive outcomes, even among interventions broadly classified as exercise games or movement-based games? The heterogeneity in game design, interaction paradigms, and cognitive engagement strategies may contribute meaningfully to the observed variability in treatment effects.

Given these considerations, the following sections discuss the implications of our findings for two key stakeholder groups: (1) designers and developers creating future exercise game interventions, and (2) researchers conducting investigations with exercise game platforms in community settings.

4.4.1 Implications for exercise game designers and developers

Our findings have several important implications for exercise game developers and designers. First, prioritizing multicomponent exercise is crucial, as our results strongly suggest that incorporating both aerobic and non-aerobic elements into exercise games is essential for maximizing cognitive benefits. Developers should focus on creating games that elevate heart rate, provide resistance exercises, and engage players cognitively, as resistance activities not only enhance cognitive function in older adults (Li et al., 2018; Huang et al., 2022) but also help maintain and build muscle mass, supporting motor functions (Mayer et al., 2011) and facilitating participation in repetitive exercises, such as aerobics. Second, while exercise games have shown broad cognitive benefits, targeting specific cognitive domains—particularly memory and processing speed—could be particularly advantageous, and designing games with challenges that explicitly address these domains might enhance their impact. Third, recognizing the differential effects of exercise games on different populations, developers could create tailored experiences for healthy older adults, individuals with MCI, and those with dementia to optimize outcomes for each group. Finally, making in-game data collection available to researchers through special access would enable researchers to have a better understanding of individual progress within exercise games and how that relates to cognitive outcomes, thereby advancing both practical application and scientific knowledge in this field.

4.4.2 Implications for researchers

Our findings highlight several important directions for future research. First, future research should prioritize studies investigating the effects of aerobic and multicomponent exercise games on cognition in older adults, as these intervention types show the most promise. Second, to better understand the underlying mechanisms of cognitive improvement, research should delve deeper into cognitive mechanisms by exploring the neurophysiological changes associated with exercise game interventions (e.g., using techniques such as EEG, ERPs, MRI, and fMRI), and future research should also explore whether specific types of memory, such as working memory, procedural memory, declarative memory, and long-term memory, show the more substantial benefits of exercise games. Third, addressing assessment limitations is essential for advancing the field; developing and validating more sensitive and specific cognitive assessment tools for use in community-based settings is critical, and additionally, researchers need tools that can deploy the assessments, ideally, into the game platform by collaborating with game designers and developers to better understand what happens during gameplay and how it is related to cognitive outcomes. Fourth, longitudinal studies are needed to investigate long-term effects by examining the long-term cognitive benefits of exercise games and to determine if these effects persist after completing the intervention. Finally, standardizing exercise game reporting is crucial—researchers should be encouraged to provide detailed descriptions of exercise games, including the average heart rate range during exercise, game mechanics, difficulty levels, and feedback mechanisms. It is important to observe whether different modes of exercise games (e.g., target-hitting, strategy, and narrative) may yield varying cognitive benefits, with each type potentially showing the largest gains in its corresponding cognitive domain (e.g., attention, reasoning, and verbal skills respectively), as this standardization will facilitate the categorization of exercise types, comparisons, and meta-analyses.

5 Limitations

This meta-analysis has several limitations related to both the included evidence and the review methodology. First, the small number of eligible studies (n = 10) and variability in sample sizes across these studies limited the statistical power and precision of pooled estimates; this also constrains the reliability of publication bias assessments, as funnel plot-based methods and Egger's regression are known to be underpowered when fewer than 10–15 studies are available, increasing the risk of false-positive and false-negative results. Therefore, publication bias assessments should be interpreted with considerable caution, and the potential for unpublished null findings to alter our conclusions cannot be ruled out. Additionally, our ≥12-week intervention criterion, while grounded in dose-response literature suggesting that longer interventions produce more robust cognitive benefits (Sanders et al., 2019), may have excluded shorter yet potentially informative studies. Some cognitive and physical training programs of 6-8 weeks duration have demonstrated measurable cognitive effects in specific domains (Smith et al., 2009), and excluding these studies may have limited our ability to examine dose-response relationships and reduced the overall evidence base.

Consequently, specific cognitive domains were underrepresented, with processing speed and verbal fluency each having insufficient studies (k = 1–2) to conduct meaningful meta-analytic synthesis or assess heterogeneity. The preponderance of studies focused on cognitively healthy older adults, limiting the generalizability of findings to clinical populations, particularly people with dementia where only three studies were available.

Second, substantial methodological heterogeneity existed across the included studies. Cognitive outcomes were measured using diverse assessment instruments with varying psychometric properties, making direct comparisons challenging. Exergaming interventions varied considerably in game type (e.g., dancing, sports, and balance games), exercise intensity, session duration, and total intervention length, which may have contributed to differential effects that could not be fully disentangled through subgroup analyses due to limited study numbers. Additionally, control group conditions were inconsistent, ranging from passive waitlist controls to active control interventions, potentially influencing the magnitude of observed effects.

Despite these limitations, the generally low-to-moderate heterogeneity between studies (I² = 0%–40.5% for most analyses) and low weighted risk of bias across methodological domains suggest reasonable consistency in the observed effects. Future research would benefit from standardized cognitive assessment batteries, detailed reporting of intervention parameters, larger and more diverse samples including underrepresented clinical populations, and exploration of dose-response relationships to establish optimal exergaming protocols for cognitive enhancement in older adults.

6 Conclusion

This systematic review and meta-analysis examined the effectiveness of exergaming interventions on cognitive outcomes in community-dwelling older adults, including those with mild cognitive impairment and dementia. The findings demonstrate that exergaming produces small but statistically significant improvements in overall cognitive function, though the evidence base for clinical populations remains limited and inconclusive, with only two studies examining MCI and three examining dementia, neither showing statistically significant effects. Multicomponent exercise programs incorporating aerobic elements showed numerically larger, though non-significant, effects compared to other modalities, suggesting that exercise type may influence outcomes, though more research is needed to confirm these patterns. These results provide preliminary evidence supporting the potential of exergaming as a modest, accessible supplementary intervention for cognitive health in community settings. However, the small effect sizes and limited evidence in clinical populations indicate that exergaming should be considered as one component of a comprehensive approach to cognitive health rather than a standalone intervention. Further research with larger samples, standardized protocols, and systematic safety reporting is needed to establish optimal implementation strategies and determine long-term efficacy.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The authors acknowledge the use of Claude.ai (Sonnet 4.5), a generative AI assistant developed by Anthropic, for editing and grammar checking of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frdem.2026.1768487/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Diana Carolina Oviedo, Catholic University Santa Maria La Antigua, Panama

Reviewed by: Katherine Baker, Northumbria University, United Kingdom

XiaCheng Song, National University of Malaysia, Malaysia