This 4-arm, parallel design randomized clinical trial was conducted at Railway Hospital in Rawalpindi, Pakistan, from September 2021 to December 2022, with participants randomly allocated in a 1:1:1:1 ratio to one of the 4 intervention groups. The methodology follows the Consolidated Standards of Reporting Trials (CONSORT) 2010 [40] guidelines for randomized trials and the CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) checklist for reporting exergame interventions [41], with the completed checklists provided as Multimedia Appendices 1 and 2.

We have structured the description of both the intervention and control conditions using the Consensus on Exercise Reporting Template. The interventions aligned with the Helsinki Declaration of Ethical Principles for Medical Research. Participants were recruited through 2 medical camps and community advertisements, including radio announcements (FM 102.2) and printed flyers. Follow-up telephone calls were conducted to ensure participant compliance with the intervention protocol.

The participants were male and female aged 60-75 years who met the Petersen/Winblad criteria for MCI [42] and had a Montreal Cognitive Assessment (MoCA) score of 18-25 (Urdu Version) [43], consistent with more recent literature defining the MCI range, which slightly extends below the range (20-24) specified in the trial registry. Participants underwent a medical screening to confirm exercise safety. Those with controlled metabolic or cardiovascular conditions (eg, well-managed hypertension or type 2 diabetes) cleared by their physician were eligible. Additional inclusion criteria required participants to have the ability to read and write in Urdu, adequate vision for engaging in physical and exercise training, and a balance score of 20-25 (mild balance impairment category) on the Mini-Balance Evaluation Systems Test (Mini-BESTest), ensuring that participants had sufficient baseline balance capacity to safely perform exergame tasks. As the intervention involved the exergame technology, the ability to understand and follow the exergame requirements was considered necessary and thus served as an implicit eligibility criterion. Participants involved in any other exercise of >150 min/week, impaired mobility, history of dizziness, vertigo, clinically significant psychiatric, other neurological, unstable or severe cardiovascular, metabolic, or orthopedic conditions were excluded [44]. At baseline, demographic information (age and sex) and clinical measures (BMI and Mini-BESTest scores) were collected for all participants before randomization. Age and sex were recorded through participant self-report and confirmed against medical records where available. Height and weight were measured using a standardized stadiometer and calibrated digital weighing scale, and BMI was calculated as weight in kilograms divided by height in meters squared (kg/m²). The Mini-BESTest was administered by trained physical therapists following standardized procedures to assess baseline balance function.

The study protocol was reviewed and approved by the Riphah University Ethical Review Committee (approval number Riphah/RCRS/REC/1954) and was registered at ClinicalTrials.gov (NCT04959383). All participants were provided with detailed information regarding the study’s objectives, procedures, potential risks, and benefits. They had the opportunity to ask questions, and written informed consent was obtained before participation, ensuring their voluntary involvement. To maintain privacy and confidentiality, all data were anonymized and deidentified before analysis. No identifiable images of participants are included in this manuscript. Participants did not receive any financial compensation for their participation. The overall process of participant enrolment, randomization, follow-up, and dropout was documented according to CONSORT guidelines, with the flow diagram presented in the Results section (Figure 1).

The sample size was calculated using G-Power software (Kiel University, DE) [45], based on a Cohen f effect size of 0.38 for MoCA [46]. Using a 95% confidence level (α=0.05) and 80% power, with 4 groups and one covariate, the minimum required sample size was 80 participants. Accounting for an expected attrition rate of 20%, a total of 97 patients were recruited.

Participants were selected through a nonprobability convenience sampling technique and randomized using a lottery method. To ensure allocation concealment, 100 sealed opaque envelopes (25 per group) containing prelabeled group assignments were prepared in advance by the principal investigator. Participants selected an envelope only after completing all preintervention assessments, preventing knowledge of allocation prior to baseline testing. The principal investigator generated the allocation sequence and prepared the envelopes, while the intervention administrator, who was not involved in baseline assessments, opened the envelopes and assigned participants to their respective groups. The intervention providers were not blinded due to the nature of the study [47]. Participants were blinded to the study hypothesis and were not informed about the other intervention groups. To ensure blinding, assessments were conducted by a single trained physical therapist who was not involved in the intervention. The assessor was unaware of the group allocation.

The balance-based exergame training used in this study was developed through a structured, theory-driven process. A prior EEG-based study demonstrated that each difficulty level elicited distinct acute cognitive demands, thereby validating the capacity of the training framework to deliver graded cognitive load. These findings provided the rationale for stratifying participants by difficulty level in the present trial [38]. The intervention design was guided by the Design, Dynamics, and Experience framework, which helped structure gameplay elements across design mechanics, real-time interactions, and user experience [48]. Although end users were not directly involved in development, the Design, Dynamics, and Experience framework supported the integration of meaningful cognitive-motor challenges. Additionally, the intervention was grounded in core neuroplasticity principles, specificity, repetition, intensity, and salience, and aligned with motor learning frameworks to promote cognitive-motor engagement and adaptive outcomes [49]. The manipulation of cognitive task variables, such as dual-task components, attention demands, and response inhibition, was guided by the theoretical foundation of the Challenge Point Framework and motor learning theory, which advocate for optimizing task difficulty relative to individual capacity to enhance learning and neuroplastic adaptation [50].

Weight-shifting was selected as the core motor task due to its safety and feasibility for individuals with MCI, its role as a foundational balance skill essential for gait initiation, and its suitability for precise cognitive-motor manipulation without altering the primary motor task. This choice enabled controlled, progressive targeting of both balance and gait-related postural control in a clinically safe, evidence-informed manner. Details of the intervention parameters and training protocol are mentioned in Table 1, and intervention specifications are provided in Table 2.

The control condition consisted of motor-cognitive exergaming using a recreational exergame (Wii Fit), which was selected as an active comparator rather than a passive control because it is validated for improving balance and incorporates cognitive-motor engagement [54]. This alignment reflects the current classification framework that distinguishes exergame-based motor-cognitive training (serious exergame), designed for rehabilitation purposes, from motor-cognitive exergaming (Wii Fit), which is primarily developed for entertainment. The use of Wii Fit allowed comparison of a purposefully designed, theory-driven serious exergame intervention with a widely available, recreational exergame, while ensuring participant engagement, reducing dropout risk, and providing an appropriate control group intervention.

Balance and gait outcomes were assessed by the evaluator using standardized clinical tools and a validated mobile app. The Mini-BESTest was administered as a screening measure of balance prior to enrolment [55]. Outcome measures included the timed up and go (TUG), cognitive time up and go (C-TUG), and the Gait & Balance (G&B) smartphone app. These instruments were selected for their well-established psychometric properties and clinical relevance in older adults [56-58]. Together, they provide a comprehensive evaluation of both static and dynamic balance, as well as spatiotemporal gait parameters, enabling a multidimensional assessment of mobility and fall risk.

Few deviations from the initially registered clinical trial protocol were made to enhance methodological rigor and study quality.

Data analysis was performed using SPSS Statistics for Windows (version 26; IBM Corp). The Kolmogorov-Smirnov test was conducted to assess normality. For TUG and C-TUG, the data were normally distributed. However, the data were nonnormal for all of the variables of the G&B mobile app except for HF walking speed, HT walking speed, and HT step time. Log transformation was applied to normalize the data with a nonnormal distribution. The Box-M test was used to assess the assumption of homogeneity, with results >0.05 for all variables. A mixed-model analysis of covariance (ANCOVA) with baseline values set as covariates was used to account for potential baseline differences [65,66]. The analysis determined the main effects of both “groups” and “time” as well as interaction effects between “groups” (low, moderate, high difficulty, and Wii Fit groups) and “time” (baseline, 4 weeks, and 8 weeks). The assumption of sphericity was assessed through Mauchley test of sphericity. Post-hoc analysis with Bonferroni correction was applied to adjust the significance level for multiple comparisons and reduce the risk of type I error [66]. ANCOVA-adjusted mean differences between groups were reported to account for baseline scores, and Bonferroni-corrected pairwise comparisons (Bonferroni-adjusted t tests) were then conducted to identify which groups differed significantly. A statistical significance level of 0.05 was established to test the null hypothesis, and the effect size was computed using η²ₚ (partial eta-squared) interpreted according to conventional benchmarks (η²ₚ; small=0.01, medium=0.06, large=0.14) [67].

The calculated sample size was 80; however, to account for potential attrition, a total of 97 participants were initially recruited for the study. After 4 weeks (mid assessment), 92 participants completed the intervention (low difficulty: n=23; moderate difficulty: n=22; high difficulty: n=24; Wii Fit group: n=23), and by 8 weeks, 87 participants had completed the intervention (low difficulty: n=21; moderate difficulty: n=21; high difficulty: n=23; Wii Fit: n=22). This represented an overall attrition rate of 10%, well within the <20% threshold commonly cited for randomized controlled trials [68]. Completion adherence was 91.3% in the low-difficulty group, 87.5% in the moderate-difficulty group, 92.0% in the high-difficulty group, and 88.0% in the control group, indicating strong participant engagement across all arms. All participants randomized were included in the primary analysis according to their original assigned groups (intention-to-treat). Missing data due to dropout (10%) were addressed using the mixed model method, ensuring unbiased estimation and preservation of randomization integrity. The CONSORT flow diagram (Figure 2) illustrates participant recruitment, allocation, follow-up, and analysis, and demographic characteristics are provided in Table 3.

To the best of our knowledge, the impact of balance-based exergame training involving graded cognitive-motor challenges on gait and balance functions in older adults with MCI has not been thoroughly assessed. The study showed significant improvement in dynamic balance (measured using TUG and C-TUG tests), as well as in AP steadiness with FirmEO condition. Improvement was also noted in gait parameters, including step time and step time variability under HF and HT conditions. The training was designed to enhance controlled weight-shifting in an exergaming environment, promoting both static and dynamic balance, alongside cognitive improvement. The effect of this intervention in enhancing various executive functions has already been reported, with more pronounced improvement in the high difficulty group [39].

This study identified significant improvement in TUG and C-TUG tests among participants in the low and high-difficulty groups, while the moderate-difficulty group showed comparatively less progress. This pattern can be understood through several theoretical frameworks in motor learning and rehabilitation science.

The differing outcomes across difficulty levels reflect that cognitive and motor learning have distinct optimal challenge thresholds. In the low-difficulty group, tasks were cognitively undemanding, allowing participants to allocate attentional resources to motor refinement, resulting in significant gains in balance and mobility tasks but little measurable cognitive improvement. This aligns with Fitts and Posner cognitive stage of motor learning [69], where reduced cognitive load supports the acquisition of correct movement patterns. In contrast, the moderate-difficulty group may have occupied a zone of challenge that is cognitively demanding enough to divide attention and increase mental load, but not sufficiently intense to trigger the deeper engagement and neuroplastic motor adaptations. By contrast, the high-difficulty group provided intense cognitive-motor demands and error-driven feedback, fulfilling key principles of experience-dependent neuroplasticity (specificity, salience, and challenge) [49]. This higher level of demand required participants to integrate motor planning, postural control, and rapid corrective adjustments, translating into improvements across both cognitive parameters and speed-related gait and fall-risk outcomes.

These findings align with previous research emphasizing that the cognitively challenging training enhanced balance performance [70,71]. Virtual reality–based cognitive training has been shown to improve cognitive function and performance in both TUG [25,70,72] and C-TUG tests [70]. Exergaming facilitates sensorimotor integration by enhancing the brain’s ability to process and integrate visual, vestibular, and proprioceptive inputs, leading to more efficient motor control [73]. The real-time visual feedback provided during exergaming aids in correct postural alignment and movement patterns, resulting in better coordination of muscle groups for specific motor tasks and ultimately contributing to improving TUG performance [72]. Moreover, exergaming on unstable surfaces introduces simultaneous multitasking, which enhances the brain’s capacity to manage motor and cognitive tasks concurrently, an essential aspect of executive functions. Regular engagement in these graded cognitive-motor challenges has been shown to improve dual-task performance, as observed in the C-TUG [74].

Interestingly, no significant differences in TUG and C-TUG improvements were observed across different difficulty groups overall. However, the group comparison indicated that the high-difficulty training group showed a better TUG performance than the Wii Fit group, suggesting that higher cognitive demands may be needed to effectively enhance mobility outcomes. In contrast, for C-TUG, a significant improvement was found between the low-difficulty and Wii Fit groups, suggesting that dual-task mobility improved most when participants trained under conditions of manageable cognitive load. According to cognitive load theory [75], tasks that do not overwhelm working memory allow attentional resources to be shared across motor and cognitive domains, supporting dual-task integration. By contrast, moderate and high difficulty tasks may have exceeded participants’ cognitive capacity, overloaded executive function, and limited gains in dual-task walking performance. An additional explanation may be related to attentional prioritization, whereby participants exposed to higher difficulty levels may have prioritized motor performance during training at the expense of cognitive performance. This would explain why improvements emerged primarily in single-task TUG but not in C-TUG performance, consistent with previous findings on prioritization effects in dual-task performance [76].

A significant group × time interaction effect was observed for AP steadiness under the FirmEO condition, indicating a marked difference in AP stability among various difficulty levels of exergame balance training. AP movements, such as forward and backward shifts, are generally easier to control, as they engage larger muscle groups, including quadriceps and hamstrings [77]. In contrast, ML movements, which involve side-to-side shifts, require more intricate control from smaller stabilizing muscles, along with greater reliance on proprioceptive and vestibular systems [77]. Additionally, AP steadiness tends to improve more readily due to the anatomical alignment of the foot, which is optimized for efficient movement control in the forward-backward plane. This finding is consistent with previous research, which suggests VR training significantly enhances AP stability in adults with MCI [78]. However, a study with contradictory findings reported great improvement in ML stability compared with AP stability [79]. This difference may be explained by the higher initial instability in the ML direction observed in that study, along with the training conducted on a stable surface that emphasized side-to-side movements, which could account for the greater enhancement of ML stability relative to AP stability.

A meta-analysis revealed that individuals with MCI demonstrate greater postural sway while standing with eyes open (EO) compared with healthy controls. This may be attributed to impairments in visual processing and visuospatial integration; when these impairments are more serious, balance deficits become more pronounced, since vision is essential for achieving a stable posture [7]. The significant improvement in AP steadiness with EO condition observed in this study underscores the importance of visual feedback in postural control. With EO, participants continuously receive visual information about their body’s position, enabling precise and timely postural adjustments relative to the environment. This visual input also enhances the integration of sensory information from different systems, facilitating more rapid learning and contributing to the observed improvement in AP balance during exergaming [80]. On the other hand, in the EC condition, individuals must rely heavily on proprioceptive and vestibular inputs to maintain balance. In individuals with MCI, these systems may be less effective due to cognitive decline affecting sensory integration and processing, further challenging postural control. This impaired integration can make it more challenging to maintain AP and ML steadiness with eyes closed [80].

The post-hoc analysis demonstrated no significant difference between difficulty groups. Static steadiness tasks, which primarily assess balance in a fixed position, may not be sensitive enough to detect improvements in balance during less challenging tasks, as these tasks do not fully reflect the dynamic balance demands required in higher difficulty levels. This may explain the lack of significant differences in EC steadiness across difficulty levels. However, participants in the low, moderate, and high-difficulty groups showed significant improvement in AP steadiness under the EO condition compared with the Wii Fit group. These results suggest that higher difficulty exergaming effectively improves AP steadiness, likely due to the increased physical and cognitive demands of such exercises, which promote greater focus, coordination, and overall balance stability [31].

A significant difference in improvement was observed among the groups after 8 weeks of intervention in step time and step time variability during waking under the HF conditions. Step time variability also differs across the groups under the HT condition, as indicated by mixed ANCOVA analysis. Gait is a multifaceted task requiring substantial attention, which involves physical capacity, balance, and cognitive control, including attention allocation, inhibition of distractions, and motor planning [81]. Integrating cognitive interventions, such as dual-tasking exergaming with cognitively challenging activities, may further enhance gait performance [33]. The literature supports these findings, showing significant improvement in gait speed with exergaming [31,33], and gait speed is widely recognized as one of the primary gait parameters positively influenced by exergaming [33]. Furthermore, step time variability serves as an important indicator of gait stability [33], suggesting that exergaming not only improves speed but also contributes to greater stability and control during walking.

The post-hoc analysis revealed differential improvements across groups in specific walking parameters under dual-task conditions. Notably, the high-difficulty condition yielded greater reductions in step time variability (in HF) and significantly faster walking speed (HF), as well as lower step time (HT), compared with the Wii Fit group. These results indicate that increased task challenge, combining cognitive load and physical demand, selectively enhanced motor control, consistency, and gait efficiency.

This aligns with findings from a recent randomized controlled trial and a meta-analysis, both of which reported that exergames incorporating greater postural and cognitive demands yield more pronounced improvements in gait and balance [32,82].

Furthermore, studies emphasize that exergames integrating cognitive-motor dual-tasking—especially in those with higher difficulty levels- produce gains in executive function and attentional control that mediate gait improvements [29]. Neurophysiological research has shown that increased frontal theta activity during higher-demand exergames reflects enhanced neural engagement of cognitive control networks essential to coordinate timing and balance [83]. These neurocognitive demands likely improved executive-motor integration, thereby supporting faster and more stable walking under HF/HT conditions [84].

Building on these findings, a differentiated pattern emerged across difficulty levels. The low- and high-difficulty exergaming groups achieved the most notable gains, though likely through different mechanisms. Low-difficulty tasks appeared to provide an accessible point for motor practice by reducing cognitive demands and minimizing dual-task interference, thereby enabling participants to consolidate core postural strategies and movement patterns [85,86]. By contrast, high-difficulty tasks placed greater demands on attention, motor planning, and adaptive control, which evidence suggests can stimulate neuromuscular activation, sensorimotor integration, and neuroplastic changes leading to improvements in gait variability and walking speed [32]. The moderate-difficulty condition, however, seemed to sit in a “grey zone,” neither simple enough to free attentional resources for basic motor learning nor challenging enough to drive significant motor adaptation [87]. Together, these findings highlight the importance of carefully selecting the exergame task difficulty, starting with manageable tasks to consolidate fundamentals, then advancing to higher-demand tasks to trigger more complex motor adaptations and maximize therapeutic benefit [32,87]. Although task difficulty in this study was not individualized, the observed response patterns across groups are consistent with the principles of the optimal challenge point framework, which suggests that different levels of task demand may elicit distinct motor and cognitive adaptations [88].

The study findings suggest that graded cognitive-motor challenging exergames may be a valuable approach to targeting balance and gait deficits, emphasizing the potential for these training programs to optimize functional outcomes. The results also suggest that incorporating balance-focused exergame training with graded difficulty levels can enhance balance and gait performance in older adults with MCI, and potentially extend to other populations experiencing cognitive-motor impairments, such as individuals with other neurological conditions. These potential benefits highlight the clinical relevance and impact of this intervention [89]. Clinicians may consider incorporating such interventions alongside conventional therapy to enhance dual-task performance and potentially reduce fall risk in this vulnerable and fall-prone population [90]. Because exergame difficulty can be adjusted, therapists can tailor challenges to individual capabilities and progressively modify tasks to maintain optimal engagement, making this approach suitable for use in outpatient clinics.

The study had some limitations. First, the 8-week intervention duration may have restricted the overall impact of the training, and extending the intervention duration could potentially reveal more significant differences between various difficulty levels. Although adherence was closely monitored to ensure full exposure to the planned dose, longer interventions in future studies may reveal stronger or more sustained effects. Second, since the majority of the participants were male, the findings may not be fully generalizable to both sexes. Improvement in balance and gait differs between males and females [91], and this difference may persist when exposed to varying levels of exergaming difficulty. Recruiting a more sex-balanced sample is essential to enhance the generalizability of findings and to explore potential sex-based differences. Third, the C-TUG test used the same starting number and subtraction pattern throughout the study. This design may have introduced familiarity effects, potentially reducing the test’s sensitivity to detect subtle changes in dual-task performance and cognitive-motor interference across groups. Moreover, cognitive task performance (ie, number of correct responses) was not recorded during the C-TUG, preventing evaluation of possible dual-task trade-offs where mobility improvements could occur at the expense of cognitive accuracy or vice versa. Future studies should vary the secondary task and record accuracy to capture dual-task interference better.

Fourth, the intervention used a fixed design, although tasks were structured at low, moderate, and high levels; no progression mechanism was applied across the 24 sessions. It is therefore possible that motor or cognitive learning reduced the internal load over time, limiting the challenge necessary to trigger optimal neuroplastic adaptations. Future research should integrate progressive overload principles to maintain optimal challenge. Fifth, no objective or subjective measures of task demand or internal loading metrics (eg, perceived exertion or cognitive load) were calculated, limiting the ability to confirm whether the intended challenge levels were achieved for each individual. Future studies should incorporate these assessments to strengthen intervention validation. Six, as the trial was powered on cognition rather than motor outcomes, it is possible that some gait or balance effects were underpowered; future trials may consider sample size estimation directly based on these parameters. Seven, while the G&B app is valid for core gait measures, its reliability for step time variability and asymmetry is limited. Findings related to these variables should therefore be viewed in light of their known measurement limitations.

Future studies should test difficulty-graded exergames in broader clinical populations (eg, Parkinson disease and poststroke) and examine their long-term effects. Incorporating objective and subjective measures of task demand (eg, perceived exertion and cognitive load) will enable difficulty to be tailored to individual capacity and adjusted progressively, providing clearer insight into optimal dosing. Research should also evaluate how these interventions can be embedded into fall-prevention and healthy aging programs to maximize their impact on functional decline, independence, and accessibility across clinical, community, and home settings. The future work should also explore the integration of step-based training components alongside weight-shifting tasks to enhance gait adaptability, balance-gait synergy, and ecological validity, particularly in populations at risk of mobility decline.

In conclusion, the 8-week exergame balance training effectively improved balance and spatiotemporal gait measures, including TUG, C-TUG, and AP stability on a firm surface, as well as step time and step time variability during walking under HF and HT conditions. However, no significant differences in improvement were observed across different difficulty groups for clinical balance measures, stability, and gait parameters. Both high-difficulty and low-difficulty groups showed improvement in stability, step time, and step length variability. Moreover, step time and walking speed with HF and HT significantly improved in the high-difficulty group. These findings suggest that exergame-based interventions with varying difficulty levels may enhance balance and gait parameters in older adults with MCI.