Effectiveness of an exergame-based training program on physical and cognitive function in older adults with cognitive impairment: a randomized controlled trial in rural China

Abstract

Background

Cognitive impairment is common among older adults, particularly those residing in long-term care facilities (LTCFs), where limited resources and staffing constrain access to structured rehabilitation programs. This study aims to assess the clinical effectiveness of an exergame-based training program delivered via “WarioWare: Move It!” in improving physical and cognitive functions in elderly LTCF residents.

Methods

A randomized controlled trial was conducted across multiple rural LTCFs in Shanxi Province with participants aged 65 and older. The intervention group received a 12-week exergame-based program involving motion-sensing activities, while the control group received standard care. Physical and cognitive functions were assessed using validated clinical tests, and training effects were analyzed via mixed ANOVA.

Results

A total of 226 participants were recruited, including 30 with mild dementia, 18 with moderate dementia, and 178 with mild cognitive impairment. Significant group × time interactions were found for key outcomes, including improvements in flexibility, joint mobility, balance, hand dexterity, and cognitive performance (all P < 0.05). Notable improvements were observed in sit and reach distance, shoulder and elbow range of motion, walking performance, and MoCA scores.

Conclusion

This low-cost, exergame-based intervention significantly enhanced both physical and cognitive functions in rural LTCF residents, demonstrating its feasibility and scalability for elderly rehabilitation in resource-limited settings.

Trial registration

ClinicalTrials.gov NCT06717971; December 10, 2024; (https://clinicaltrials.gov/study/NCT06717971).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-025-06341-6.

Introduction

As the global population ages, cognitive health has become a significant public health issue. Among age-related cognitive disorders, dementia is characterized by the decline in memory, executive function, attention, and language abilities, which severely impacts daily functioning and independence [1]. However, the pathological changes associated with dementia often begin several years before clinical symptoms appear. Mild cognitive impairment (MCI) is commonly regarded as an intermediate stage between normal aging and dementia, and has been identified as a key target for early detection and prevention strategies. Studies show that approximately 5–13% of individuals with MCI progress to dementia each year [2], especially when risk factors such as advanced age, genetic susceptibility, and a sedentary lifestyle are present. Given the lack of a definitive cure for dementia and the increasing burden it imposes on individuals, families, and healthcare systems, early prevention and the preservation of cognitive function are of critical importance.

In recent years, exergaming has emerged as a novel intervention that combines physical activity with interactive gaming elements, gaining increasing adoption in training and rehabilitation programs for older adults [3, 4]. By leveraging immersive virtual environments and interactive gameplay, exergames offer a dynamic platform that not only promotes physical engagement but also facilitates cognitive enhancement, making exercise more enjoyable and less monotonous for elderly participants. Research has demonstrated that various commercial gaming platforms and bespoke virtual reality games have a substantial impact on cognitive function, particularly in domains such as memory, attention, and executive control [5–7]. In addition, exergames create an engaging, gamified environment for elderly users, which can improve participation rates and adherence to training. The game-based training approach not only effectively stimulates cognitive function but also provides a convenient exercise pathway for older adults with mobility impairments who may be reluctant or unable to engage in regular outdoor physical activities, thus overcoming barriers such as adverse weather or limited accessibility [8, 9].

Despite growing evidence supporting the benefits of exergames for community-dwelling older adults [10, 11], their application in rural long-term care facilities (LTCFs) remains underexplored. Rural LTCFs often face severe resource limitations—including shortages of trained caregivers, insufficient equipment, and limited access to structured rehabilitation programs [12]. These constraints hinder the implementation of conventional exercise regimens, leaving elderly residents without effective means of maintaining physical and cognitive health. Moreover, older adults in institutional settings frequently report boredom with repetitive routines and exhibit low intrinsic motivation to engage in traditional exercises, especially when deprived of novelty or feedback [13]. In this context, gamified interventions such as exergames may offer an engaging alternative, particularly if they are low-cost, easy to implement, and adaptable to varying functional levels.

However, for exergame-based interventions to be truly effective in LTCF contexts, it is not enough for them to be merely engaging—they must also produce clinically relevant improvements in core physical functions that support independence. While prior exergame research has shown improvements in general health outcomes, cognitive function, and quality of life [14, 15], relatively few studies have assessed targeted physical functions (e.g., flexibility, joint mobility, coordination, and hand dexterity), especially in LTCF populations. These physical capabilities are critical for maintaining autonomy in later life. For example, diminished flexibility and joint mobility can hinder activities of daily living [16, 17], while reduced coordination is a major risk factor for falls [18], and decreased hand dexterity can impair self-care and social participation [19]. Therefore, it is essential that exergame interventions target these core physical functions to maximize their benefits for elderly individuals in LTCFs.

To address these gaps, this study introduces a novel exergaming protocol that leverages the unique properties of WarioWare: Move It, a Nintendo Switch game specifically suited for dynamic, multi-domain physical training. In contrast to previous Wii-based interventions that typically relied on coarse motion tracking and repetitive sports-themed content, WarioWare: Move It delivers over 200 diverse and fast-paced microgames that promote spontaneous, full-body engagement. These microgames require quick transitions between tasks such as reaching, crouching, twisting, and balancing, thereby engaging critical physical domains including flexibility, joint mobility, coordination, and postural control within brief, cognitively stimulating intervals. The inherently unpredictable and humorous nature of the game reduces habituation and sustains attention, addressing a common limitation of monotonous exergame routines. Importantly, the intervention incorporates manual adjustment of game parameters (e.g., movement range, pace, and sequence order), allowing individualized calibration of challenge and intensity. Building on these features, the present study evaluates the feasibility and clinical effectiveness of WarioWare: Move It in improving both physical functions (e.g., flexibility, joint mobility, coordination, and hand dexterity) and cognitive performance among older adults with cognitive impairment in rural long-term care facilities. By targeting specific domains of functional decline through an engaging, adaptive, and scalable platform, this research aims to inform geriatric rehabilitation practices and guide the development of elderly-friendly exergame systems in resource-constrained settings.

Methods

Ethical considerations

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Biomedical Ethics Review Committee of Taiyuan University of Technology (Approval No. 20230641). Given the inclusion of older adults with suspected or diagnosed cognitive impairment, informed consent was obtained for all participants prior to their enrollment in the study. Individuals with sufficient cognitive capacity provided written informed consent independently. For participants with impaired decision-making ability, written consent was obtained from their legally authorized representatives (e.g., family members or appointed guardians), in accordance with institutional review board guidelines. The consent procedure was conducted by trained research staff in a quiet and private setting. A simplified explanation of the study’s purpose and procedures was initially provided to each participant. If limited comprehension or uncertainty was observed, the process was immediately redirected to their legal guardian, who then completed the informed consent process on their behalf. As a token of appreciation, eligible participants received a compensation of USD 10.

Participants

In this study, we recruited 226 participants (30 patients with mild dementia, 18 with moderate dementia, and 178 with mild cognitive impairment) from LTCFs, such as rural daycare centers and nursing homes in Shanxi Province, China. Participant selection followed a purposive sampling method [20], with the entire recruitment process rigorously supervised by experienced neurologists. Inclusion criteria were as follows: (1) normal or corrected-to-normal hearing and vision; (2) age over 65; (3) residence in a senior care facility or a minimum stay of one month in such a facility; (4) completion of the Global Deterioration Scale (GDS), and the Chinese version of the Montreal Cognitive Assessment (MoCA), with the capacity to communicate effectively; (5) ability to engage in moderate physical activity without physical disability; (6) absence of severe depressive symptoms or other neurological disorders (e.g., stroke, dizziness, epilepsy); and (7) provision of informed consent by the participant or their guardian.

All participants underwent comprehensive clinical evaluations conducted by experienced neurologists. These assessments included detailed medical histories, physical examinations, structural neuroimaging and standardized cognitive function testing. Structural neuroimaging, predominantly MRI, revealed stage-specific cerebral changes corresponding to different levels of cognitive impairment. In individuals diagnosed with mild dementia, MRI scans demonstrated mild medial temporal lobe atrophy (MTA), characterized by subtle hippocampal and cortical thinning, minimal ventricular enlargement, and mild white matter hyperintensities (WMH) primarily located in the periventricular and deep white matter regions. In contrast, individuals with moderate dementia exhibited more pronounced neurodegenerative changes, including marked temporal and frontal lobe atrophy, higher MTA scores (2–3), widened cortical sulci, enlarged lateral ventricles (particularly in the temporal horns), and moderate to severe WMH. Participants with mild cognitive impairment did not show overt structural abnormalities on MRI or CT. In addition, cognitive function severity was further classified using the GDS for participants with dementia, yielding an average GDS score of 3.65 and an average symptom duration of 4.2 years. For individuals diagnosed with MCI, the MoCA was administered, with all scores falling consistently below the clinical cutoff of 26. Table 1 summarizes the clinical and demographic characteristics of participants allocated to the intervention and control groups.

Table 1.

Participant demographic and baseline characteristics

Characteristics (units)	Intervention Group (n = 113)	Control Group (n = 113)	P-value (t-test/chi-square)
Age(years), mean(SD)	73.12(4.312)	72.29(4.852)	0.596
Gender, n(%)			0.348
Woman	67(59.3)	60(53.1)
Man	46(40.7)	53(46.9)
Hand preference, n(%)			0.703
Left	15(13.3)	17(15.0)
Right	98(86.7)	96(85.0)
Exercise habit, n(%)			0.584
Yes	41(36.3)	45(39.8)
No	72(63.7)	68(60.2)
Education (years), mean (SD)	6.47(3.727)	6.29(4.154)	0.569
Hours of sleep (hours), mean (SD)	6.47(1.334)	6.29(1.211)	0.901
Smart device use (years), mean (SD)	5.96(1.544)	5.47(1.881)	0.283

Abbreviations: SD Standard Deviation

P-values were derived from independent-samples t-test for continuous variables and chi-square test for categorical variables 

Randomization and blinding

A stratified randomization procedure was employed in this study to eliminate selection bias and ensure group equivalence across varying levels of cognitive impairment. Participants were first categorized into three subgroups based on cognitive status as determined by standardized clinical assessment: MCI (n = 178), mild dementia (n = 30), and moderate dementia (n = 18). Within each subgroup, participants were then randomly assigned to either the intervention or control group using a 1:1 allocation ratio. The randomization was conducted using a double-anonymized envelope-drawing method to maintain allocation concealment and ensure procedural rigor:

• All eligible participants were assigned a unique identification number.

• IDs were placed into opaque envelopes stratified by cognitive status.

• Envelopes within each cognitive subgroup were shuffled, and researchers wearing blindfolds randomly selected half to allocate to the intervention group; the remainder were assigned to the control group.

Post-allocation, the intervention group received a 12-week exergame-based training program, while the control group received standard care, with both groups matched for total activity duration.

Control group standard care program

Participants in the control group received a structured standard care program based on typical practices within long-term care facilities, designed to support both physical and mental health through regular group-based activities. The program included customized physical exercise sessions, horticultural activities, and low-intensity group engagements suitable for older adults. Physical exercises were carefully modified to accommodate the diverse mobility levels of participants, ensuring that all activities could be performed safely and comfortably while encouraging consistent movement. The horticultural component involved interactive gardening tasks, such as planting and plant care, which provided sensory stimulation and supported emotional well-being. In addition, sedentary activities (e.g., board games and other table-based interactions) were incorporated to promote socialization and maintain cognitive engagement in a relaxed setting. All sessions were facilitated by trained therapists, who provided personalized support and ensured the appropriateness of each activity. The program was delivered twice weekly, with each session lasting 60 min over a 12-week intervention period.

Intervention group training program

The training program for the intervention group utilized the Nintendo Switch game WarioWare: Move It, as shown in Fig. 1. All sessions were conducted in a quiet activity room, where participants sat approximately 1.5 m from a large display screen and held one Joy-Con controller in each hand. Each session followed a standardized structure across all participants, consisting of a curated sequence of mini-games designed to elicit responses across the full range of targeted physical domains, including flexibility, joint mobility, coordination, and balance. For example, tasks requiring forward reaching, crouching, or dynamic leaning (e.g., simulating pumping air or swatting at objects) were included specifically to engage hamstring flexibility and trunk rotation, corresponding to lower-limb flexibility assessments such as the Sit and Reach test. While the overarching program content was consistent, in-game parameters such as pace, movement range, and game selection order were individually adjusted by trained staff based on each participant’s physical tolerance, cognitive status, and baseline ability. These adaptations ensured that each participant engaged in tasks with sufficient intensity and functional relevance to stimulate progress in the targeted outcome domains. Importantly, care was taken to include exercises targeting both gross and fine motor responses across sessions for all participants, thus ensuring outcome-aligned exposure regardless of individual tailoring.

Fig. 1.

Training scenarios and gameplay modes of the Nintendo Switch “WarioWare: Move It” exergame used in the intervention group

During gameplay, participants followed animated demonstrations and verbal cues to perform rapid motor actions involving the upper limbs, trunk, and lower extremities. Movements included arm elevation, side lunges, step shifts, shoulder rotations, crouching, and reactive turning. To promote sustained engagement, the system provided real-time audiovisual feedback (e.g., success indicators, score displays, and sound effects). Each 60-minute session was administered twice weekly over 12 weeks and comprised three 10-minute active game segments, interspersed with 1–5 min rest periods. Staff continuously monitored participants, providing encouragement, postural correction, and adaptive assistance to ensure both safety and optimal engagement across all target domains.

Outcomes measured

All participants underwent three assessments at distinct time points. The initial assessment occurred at baseline prior to randomization. Subsequent evaluations were conducted at the end of 12 weeks (post-test) and six months after the intervention concluded (after 6 months). To minimize measurement bias, all cognitive and physical evaluations were administered by trained assessors who were blinded to group allocation. These assessors were not involved in randomization or intervention delivery, and assessment schedules were managed by an independent coordinator to maintain allocation concealment. Participants were instructed not to disclose their group assignment during evaluations. The primary outcomes of this study focus on detecting intervention effects using indicators related to physical flexibility, joint range of motion, motor coordination, hand dexterity and cognitive abilities, as detailed below.

In this study, physical flexibility assessments involve several standardized tests. First, the Sit and Reach test measures the participant’s ability to stretch forward while seated with legs extended and heels fixed [21, 22]. The farthest distance reached with outstretched hands toward the toes is recorded, with the gap between the fingertips and toes serving as the flexibility index. The Shoulder Flexibility test assesses flexibility by measuring the distance between the hands when clasped behind the back [23, 24]. Participants with higher flexibility achieve either easy hand clasping or shorter hand-to-hand distance. The Trunk Rotation Flexibility test requires participants to stand with feet together, and arms extended forward, rotating the torso left and right around the waist [25]. The maximum angle reached on each side indicates flexibility, with more prominent angles signifying greater flexibility.

Subsequently, the joint range of motion reflects the extent of freedom a joint possesses within a specified range of movement. Observing changes in the range of motion across different directions in the intervention group makes it possible to determine whether significant improvements occur in particular directions (e.g., flexion or internal rotation), thus providing targeted feedback for the intervention. This study’s joint range of motion assessments includes shoulder and elbow range of motion tests. The Shoulder Range of Motion test measures the maximum range in four directions (e.g., flexion, extension, abduction, and adduction) [26, 27]. In contrast, the Elbow Range of Motion test assesses the degree of elbow flexion [28].

Additionally, this study evaluates motor coordination, defined as the body’s ability to synchronize various parts during complex physical tasks. The assessment methods include the Figure of Eight Walk [29, 30] and Standing Balance test [31, 32]. In the Figure of Eight Walk test, participants are instructed to walk in a figure-eight pattern around two markers placed on the ground, covering a total distance of 10 m. The time taken to complete this task is recorded as a measure of coordination. The Standing Balance test requires participants to stand with their feet together, and their eyes closed, aiming to maintain balance, with the duration of balance maintained being measured as an indicator of stability.

Similarly, hand dexterity reflects an individual’s coordination, agility, and reaction speed in performing fine motor tasks, particularly when required to complete repetitive, high-precision tasks within a limited timeframe. Hand dexterity was assessed using the Box and Block Test (BBT) [33, 34], where participants were instructed to use their dominant hand to transfer blocks from one compartment to another within a wooden box over 60 s, recording the total number of transfers completed within the specified time. The BBT has demonstrated high test–retest reliability, with intraclass correlation coefficients ranging from 0.89 to 0.97 in older adults with upper limb impairments [34].

Finally, cognitive function was evaluated using the Cognitive Abilities Screening Instrument (CASI) [35], and MoCA scales [36]. The CASI is a comprehensive multidomain tool encompassing nine cognitive domains, including long-term memory, short-term memory, attention, mental manipulation, orientation, abstraction and judgment, language, visuoconstruction, and category fluency. Its construct validity has been supported by second-order confirmatory factor analysis, which confirmed that domain scores reliably load onto a general cognitive factor, justifying the use of the total score as a global cognition index [37]. The CASI also exhibits excellent test–retest reliability in dementia populations, with an intraclass correlation coefficient (ICC) of 0.97 for the total score and domain-level ICCs ranging from 0.65 to 0.92 [38]. In parallel, the MoCA was employed to assess global cognitive status, with a focus on domains such as visuospatial-executive functioning, attention, language, memory, and orientation. The MoCA has demonstrated high internal consistency, with Cronbach’s alpha reaching 0.905 in the full sample and 0.883 in clinical subgroups with cognitive impairment, confirming its psychometric robustness in both general and clinical elderly populations [39]. Detailed measurement procedures are provided in Multimedia Appendix 1.

Statistical analysis

The statistical analyses were performed using SPSS for Windows, version 21.0 (IBM Corp.). Descriptive statistics were reported as mean ± standard deviation for continuous variables and as counts with corresponding percentages for categorical variables. Prior to conducting inferential tests, assumptions of normality and homogeneity of variances were evaluated. For data meeting the criteria for normal distribution, independent-samples t-tests were utilized to compare continuous variables between groups, while chi-square tests were employed for categorical data. A two-sided P-value of less than 0.05 was considered indicative of statistical significance across all analyses.

To examine the effects of the intervention over time, a mixed-design analysis of variance (ANOVA) was employed, incorporating time as a within-subject factor and group assignment as a between-subject factor. Effect sizes were reported using eta squared (η²), with values ranging from 0 to 1 to indicate the proportion of total variance attributable to the intervention. Larger η² values suggest stronger intervention effects. When significant main effects or interactions were identified, Bonferroni-adjusted post hoc tests were conducted to explore specific group-time comparisons.

Results

Adherence, completion, and retention rates in the intervention group

To evaluate the intervention’s feasibility and participant acceptability, we analyzed session adherence, completion rates, retention, and session-level behavioral engagement across cognitive subgroups. As summarized in Table 2, the overall adherence was high. Among the 113 participants in the intervention group, 89 were categorized as having mild cognitive impairment, 15 as mild dementia, and 9 as moderate dementia. A large majority of MCI participants (n = 83, 93.3%) completed all 12 sessions, followed by those with mild dementia (n = 13, 86.7%) and moderate dementia (n = 7, 77.8%). Follow-up assessment attendance rates at six months remained relatively high across groups: MCI (87.6%), mild dementia (80.0%), and moderate dementia (66.7%). The primary reasons for attrition were unrelated to the intervention itself and included scheduling conflicts, mild illness, physical fatigue, and fluctuations in medical or cognitive status.

Table 2.

Session completion and retention rates across cognitive subgroups

Cognitive Status	N	Completed All Sessions (n, %)	Attended Follow-up (n, %)	Main Reasons for Dropout
Mild Cognitive Impairment	89	83 (93.3%)	78 (87.6%)	Scheduling conflicts, unrelated illness, other factors
Mild Dementia	15	13 (86.7%)	12 (80.0%)	Physical fatigue, minor illness, other factors
Moderate Dementia	9	7 (77.8%)	6 (66.7%)	Cognitive instability, comorbidities, low energy

In addition to attendance-based indicators, behavioral observations further reflected strong engagement with the program (Table 3). Specifically, 81.4% of participants initiated tasks voluntarily with minimal external prompting, and 85.8% sustained attention for at least 80% of session time. Notably, 76.1% exhibited visible enjoyment (e.g., smiling, laughing, or expressing verbal interest), while 70.8% engaged in social interaction with peers or facilitators during gameplay. Taken together, the favorable completion rates and behavioral observations suggest that the intervention was generally acceptable and feasible for most participants, including those with mild to moderate cognitive impairment.

Table 3.

Observed behavioral engagement characteristics

Behavior Category	Common Observations	Noted in Participants (n, %)
Task initiation	Participants voluntarily engaged with tasks at session onset.	92 (81.4%)
Sustained attention	Maintained focus during sessions (> 80% time)	97 (85.8%)
Expressed enjoyment	Smiled/laughed or verbally expressed interest	86 (76.1%)
Peer interaction	Interacted with facilitator or peers	80 (70.8%)

Baseline performance of the intervention and control groups

Prior to the initiation of the intervention, baseline assessments were conducted for both the intervention and control groups. These evaluations encompassed several key domains, including physical flexibility, joint mobility, balance, coordination, and cognitive function. As presented in Table 4, the results revealed no statistically significant differences between the groups across any of the measured variables (P > 0.05). This suggests that the two groups were well-matched in terms of physical and cognitive function at the outset of the study.

Table 4.

Data of measurement outcomes at baseline, post-test, and after 6 months post-intervention

Tests			Baseline Mean (SD)	Post-test Mean (SD)	After 6 months Mean (SD)	F test	P value	Effect size (η2)
Sit and Reach Test (centimeters)	Control group		24.92(7.826)	24.28(7.345)	24.00(8.092)	0.908a; 0.297b; 0.541c
	Intervention group		24.73(7.534)	22.50(7.776)	22.93(7.570)
	2-way ANOVA	Time	—d	—	—	35.682	< 0.001	0.314
		Group×time	—	—	—	8.484	< 0.001	0.098
Shoulder Flexibility Test (centimeters)	Control group		18.83(6.029)	18.18(5.728)	17.90(5.368)	0.785; 0.143; 0.285
	Intervention group		18.45(6.193)	16.30(5.608)	16.62(5.227)
	2-way ANOVA	Time	—	—	—	16.511	< 0.001	0.175
		Group×time	—	—	—	3.666	0.035	0.045
Trunk Rotation Flexibility Test (degrees)	Control group		103.80(19.824)	104.43(18.548)	104.05(18.026)	0.961; 0.170; 0.300
	Intervention group		103.57(21.260)	110.65(21.546)	108.60(20.853)
	2-way ANOVA	Time	—	—	—	24.011	< 0.001	0.235
		Group×time	—	—	—	17.353	< 0.001	0.182
Shoulder Range of Motion Teste (degrees)	Control group		139.70(12.584)	141.43(14.113)	140.92(13.312)	0.945; 0.010; 0.013
	Intervention group		139.53(9.814)	148.73(10.370)	147.20(8.115)
	2-way ANOVA	Time	—	—	—	36.315	< 0.001	0.318
		Group×time	—	—	—	17.655	< 0.001	0.185
Shoulder Range of Motion Testf (degrees)	Control group		114.23(12.853)	115.50(13.091)	116.00(12.922)	0.971; 0.830; 0.823
	Intervention group		114.32(11.952)	116.10(11.790)	115.40(10.916)
	2-way ANOVA	Time	—	—	—	3.164	0.055	0.039
		Group×time	—	—	—	0.395	0.635	0.005
Shoulder Range of Motion Testg (degrees)	Control group		30.88(5.774)	30.78(5.347)	31.05(5.164)	0.943; 0.180; 0.881
	Intervention group		30.78(6.612)	32.70(7.244)	31.25(6.648)
	2-way ANOVA	Time	—	—	—	4.499	0.016	0.055
		Group×time	—	—	—	6.281	0.004	0.075
Shoulder Range of Motion Testh (degrees)	Control group		22.15(4.481)	22.70(4.084)	22.48(4.403)	0.867; 0.897; 0.919
	Intervention group		22.33(4.811)	22.82(4.551)	22.58(4.314)
	2-way ANOVA	Time	—	—	—	1.774	0.177	0.022
		Group×time	—	—	—	0.009	0.984	0.001
Elbow Range of Motion Test (degrees)	Control group		125.88(9.067)	126.60(8.796)	126.10(9.353)	0.698; 0.536; 0.770
	Intervention group		125.07(9.300)	127.85(9.181)	126.67(8.113)
	2-way ANOVA	Time	—	—	—	9.261	0.001	0.106
		Group×time	—	—	—	3.298	0.049	0.041
Figure of Eight Walk Test (seconds)	Control group		45.65(8.438)	44.57(8.054)	45.40(7.585)	0.838; 0.092; 0.233
	Intervention group		45.25(9.012)	41.35(8.845)	43.22(8.577)
	2-way ANOVA	Time	—	—	—	36.038	< 0.001	0.316
		Group×time	—	—	—	11.846	< 0.001	0.132
Standing Balance Test (seconds)	Control group		19.17(6.417)	19.77(6.463)	20.05(6.089)	0.870; 0.364; 0.686
	Intervention group		19.40(5.848)	21.00(5.505)	20.57(5.449)
	2-way ANOVA	Time	—	—	—	9.059	< 0.001	0.104
		Group×time	—	—	—	1.579	0.212	0.020
Box and block Test (count)	Control group		12.38(3.484)	12.83(3.869)	13.15(2.992)	0.975; 0.327; 0.793
	Intervention group		12.35(3.759)	13.73(4.285)	13.35(3.766)
	2-way ANOVA	Time	—	—	—	18.648	< 0.001	0.193
		Group×time	—	—	—	4.016	0.022	0.049
Cognitive Abilities Screening Instrument (score)	Control group		67.05(5.702)	70.88(6.223)	71.98(6.327)	0.229; 0.002; 0.207
	Intervention group		66.45(6.093)	75.07(5.535)	72.67(5.599)
	2-way ANOVA	Time	—	—	—	234.920	< 0.001	0.751
		Group×time	—	—	—	30.914	< 0.001	0.284
Chinese Version of the MoCA Scale (score)	Control group		21.95(4.579)	22.48(4.279)	22.73(4.057)	0.722; 0.142; 0.199
	Intervention group		21.60(4.187)	23.80(3.695)	23.88(3.878)
	2-way ANOVA	Time	—	—	—	52.563	< 0.001	0.403
		Group×time	—	—	—	15.882	< 0.001	0.169

Effect sizes (η²) were interpreted as small (> 0.01), medium (> 0.06), and large (> 0.14), following conventional guidelines [40]

^aA t-test was applied for continuous variables at baseline for control and intervention groups

^bA t-test was applied for continuous variables post-test for control and intervention groups

^cA t-test was applied for continuous variables after 6 months post-intervention for control and intervention groups

^dNot applicable

^eThe maximum range of motion of the shoulder joint in flexion

^fThe maximum range of motion of the shoulder joint in extension

^gThe maximum range of motion of the shoulder joint in abduction

^hThe maximum range of motion of the shoulder joint in adduction

Comparison of physical function outcomes between intervention and control groups

To evaluate improvements in physical function, we analyzed group-by-time effects and effect sizes across multiple outcomes. As shown in Table 4, the Sit and Reach test revealed a significant improvement in flexibility over time (F = 35.682, p < 0.001, η²=0.314), with a moderate interaction effect (F = 8.484, p < 0.001, η²=0.098), indicating greater gains in the intervention group. A similar pattern was observed in the Shoulder Flexibility test, which showed a large main effect (F = 16.511, p < 0.001, η²=0.175) and a small interaction effect (F = 3.666, p = 0.035, η²=0.045). The Trunk Rotation Flexibility test revealed a significant group × time interaction (F = 17.353, p < 0.001, η²=0.182), which qualifies as a large effect. The improvement was sustained at follow-up, supporting the intervention’s long-term benefits for trunk mobility.

Furthermore, the joint range of motion also improved, particularly in shoulder flexion (F = 36.315, p < 0.001, η²=0.318) and abduction (F = 4.499, p = 0.016, η²=0.055). Group × time interaction effects were large for flexion (F = 17.655, p < 0.001, η²=0.185) and medium for abduction (F = 6.281, p = 0.004, η²=0.075), suggesting the intervention specifically enhanced dynamic upper-limb movement. For elbow flexion, both a medium time effect (F = 9.261, p < 0.001, η²=0.106) and a small interaction effect (F = 3.298, p = 0.049, η²=0.041) were observed, likely reflecting the frequent upper-limb engagement required by the game-based activities (e.g., simulated swimming or reaching).

Subsequently, motor coordination was assessed via the Figure of Eight Walk and Standing Balance tests. The walk test showed both a large time effect (F = 36.038, p < 0.001, η²=0.316) and a near-large interaction effect (F = 11.846, p < 0.001, η²=0.132), indicating meaningful gains in gait coordination. Similarly, the Standing Balance test showed a medium effect size for time (F = 9.059, p < 0.001, η²=0.104), suggesting improved postural stability in the intervention group. Finally, the Box and Block test revealed a significant increase in hand dexterity over time (F = 18.648, p < 0.001, η²=0.193) with a small but significant interaction effect (F = 4.016, p = 0.022, η²=0.049).

Across these outcomes, most interaction effects fell within the small to medium range (η²>0.01–0.14), with several large effects (η²>0.14) observed for key mobility and coordination indicators. These findings underscore the effectiveness of the exergaming intervention in improving physical function domains among older adults, with particularly strong effects in flexibility, upper-limb movement, and motor coordination.

Comparison of cognitive function outcomes between intervention and control groups

Cognitive function was evaluated by comparing changes in standardized cognitive measures (e.g., CASI, MoCA) between the intervention and control groups, as presented in Table 4. In the CASI assessment, participants’ scores significantly increased over time (F = 234.920, P < 0.001), with a notable group x time interaction effect (F = 30.914, P < 0.001), suggesting a positive impact of the intervention on cognitive function. A similar pattern was observed in the MoCA assessments. Likewise, MoCA scores demonstrated a significant time effect (F = 52.563, P < 0.001) and a group x time interaction effect (F = 15.882, P < 0.001), reinforcing the intervention’s effectiveness in enhancing various aspects of cognitive function. Notably, the effect sizes (η²) for the group x time interaction effects across the two scales were 0.284 for CASI, and 0.169 for MoCA.

In addition, we analyzed the subdomain-level changes in the MoCA and CASI scores among the 113 participants in the intervention group. As illustrated in Fig. 2, substantial proportions of participants demonstrated post-intervention gains in several cognitive subdomains. In the MoCA, improvements were most frequently observed in the Visuospatial/Executive domain (n = 84), Attention (n = 95), and Delayed Recall (n = 81), with over 70% of participants showing gains in each. In the CASI, notable improvements were found in Long-term Memory (n = 97), Short-term Memory (n = 77), Attention (n = 88), Concentration/Mental Manipulation (n = 72), and Visual Construction (n = 85), all exceeding 60% of the sample. These results indicate that the intervention particularly enhanced executive functioning, memory, and attentional processes, which are core cognitive domains affected in early-stage dementia and MCI.

Fig. 2.

Proportion of participants showing improvement in cognitive subdomains from pre- to post-intervention

Subgroup analysis of functional and cognitive gains across cognitive severity levels

To examine whether intervention effects differed by baseline cognitive function, we conducted a subgroup analysis of 113 participants in the intervention group, categorized as having mild cognitive impairment, mild dementia, or moderate dementia. Gain effects (i.e., post-intervention minus pre-intervention values) were computed for each outcome measure and statistically compared across the three groups. Significant differences in both physical and cognitive outcomes were observed, as illustrated in Fig. 3.

Fig. 3.

Between-group differences in gain effect across physical and cognitive outcomes among participants with mild cognitive impairment, mild dementia, and moderate dementia

For the Sit and Reach Test, differences emerged between MCI and moderate dementia (p = 0.044), and between mild and moderate dementia (p = 0.019). In the Shoulder Flexibility Test, significant differences were found between MCI and mild dementia (p = 0.045), and between MCI and moderate dementia (p = 0.035). The Trunk Rotation Test showed differences between MCI and moderate dementia (p = 0.020), and mild and moderate dementia (p = 0.029). For Shoulder Range of Motion (Flexion), MCI participants improved more than those with moderate dementia (p = 0.007), and a similar trend was noted between mild and moderate dementia (p = 0.040). Significant differences were also observed in the Elbow Range of Motion Test (MCI vs. moderate dementia, p = 0.032), Figure of Eight Walk Test (MCI vs. moderate dementia, p = 0.022; mild dementia vs. moderate dementia, p = 0.039), CASI Test (MCI vs. mild dementia, p = 0.038; MCI vs. moderate dementia, p = 0.005), and MoCA Test (MCI vs. moderate dementia, p = 0.047).

Overall, participants with MCI or mild dementia exhibited significantly greater functional and cognitive gains compared to those with moderate dementia, indicating that individuals with higher baseline cognitive capacity may engage more effectively with the intervention tasks and benefit more substantially from the program.

Discussion

Advantages of exergames in physical and cognitive rehabilitation for older adults

The findings of this study align with previous exergame-based interventions among community-dwelling older adults, which have consistently demonstrated benefits in enhancing physical function, cognition, and motivation through interactive gameplay [41, 42]. However, research specifically targeting LTCF populations remains limited. While earlier studies using commercial exergames like Nintendo Wii Fit or Xbox Kinect have reported moderate improvements in mobility, balance, or mood within LTCFs [43, 44], these interventions were often constrained by a narrower set of training targets, relatively basic game mechanics, or low adaptability to users with cognitive decline.

In contrast, our study using WarioWare: Move It! observed simultaneous improvements across multiple functional domains, including flexibility, joint mobility, hand dexterity, and cognitive performance. The multidomain effect can be attributed to the game’s unique design, which differs significantly from earlier Wii-based exergames in several key aspects. First, unlike balance-board–based tasks in Wii Fit or gesture-matching games in Kinect Sports, WarioWare: Move It! incorporates over 200 short microgames that require rapid response, cognitive flexibility, sensorimotor coordination, and rule-switching under time pressure. These design elements directly map onto core cognitive domains (e.g., executive control, attention regulation, and visuospatial processing) that are particularly vulnerable in MCI and early dementia. Second, the microgame structure ensures high temporal variability and novelty, reducing habituation and maintaining cognitive engagement over time. This contrasts with the repetitive or static gameplay seen in prior exergame protocols, which may lead to participant disengagement in LTCF contexts. Third, the system features intuitive motion controls via dual Joy-Con sensors and allows for a fully seated mode of play, making it more accessible for older adults with frailty, limited mobility, or fall risk—an inclusion rarely found in conventional Wii-based interventions. Furthermore, while prior LTCF studies have noted challenges with engagement, adherence, or task comprehension among residents with cognitive impairments [45], our implementation of WarioWare: Move It! demonstrated high participation and retention. It is due to its compact, humorous, and socially interactive game format, which may foster intrinsic motivation, enjoyment, and social bonding during sessions. These behavioral benefits, as reflected in observed enjoyment, sustained attention, and peer interaction, represent key facilitators of intervention feasibility and long-term adherence.

Therefore, our findings not only confirm the general feasibility and efficacy of exergames in older adult populations, but also provide novel evidence supporting their tailored use in LTCFs. By extending benefits across both physical and cognitive domains in a low-resource setting, this study contributes to a growing recognition of exergames as viable, scalable interventions in institutional elderly care.

Functional and cognitive gains attributable to exergame-based intervention

Our findings revealed statistically significant and practically meaningful improvements in multiple physical function domains following the 12-week exergame-based intervention. In particular, large effect sizes were observed for flexibility-related outcomes, such as fingertip-to-toe distance in the Sit and Reach test (η²=0.314) and shoulder range in the Shoulder Flexibility test (η²=0.175), indicating robust intervention effects. These improvements are clinically relevant, as enhanced flexibility is associated with better joint mobility, fall prevention, and greater independence in daily functioning. Age-related stiffness and reduced neuromuscular coordination [46] often impair mobility in older adults; thus, the observed gains suggest that the intervention’s repetitive, multi-directional movements may have reactivated neuromotor pathways and improved proximal joint control. Notably, trunk rotation flexibility also showed a large group × time interaction effect (η²=0.182), reflecting improvements in spinal mobility and segmental coordination—key prerequisites for safe turning, reaching, and maintaining upright posture during daily activities. These effects likely emerged through dynamic, torso-involving tasks (e.g., skiing, flying simulations) that demand axial control and core stabilization.

Improvements in joint range of motion were particularly notable in shoulder flexion and abduction, movements essential for self-care tasks such as dressing and overhead reaching [47]. The intervention group demonstrated large effect sizes in these domains, suggesting that even without external resistance or standing posture, task-relevant dynamic simulations (e.g., simulated lifting, swimming, or pushing) were sufficient to elicit neuromotor adaptation. Notably, elbow flexion improvements may reflect enhanced control of distal joints through improved motor unit recruitment and proprioceptive modulation, mechanisms known to underlie neuromuscular plasticity in older populations [16, 48].

Although lower limb motion was not the primary training target, participants exhibited statistically significant improvements in walking speed and performance on the Figure of Eight Walk test. This outcome reflects a generalization effect or motor transfer, in which training-induced improvements in trunk stability and center-of-gravity control translated into better gait performance. Games requiring dynamic leaning, torso rotation, or reaching forward (e.g., skiing, flying) may have incidentally trained balance perception and axial control, both critical for gait regulation [49, 50]. A significant increase in duration during the Standing Balance test further supports the hypothesis that neuromuscular coordination and postural control improved as a downstream effect of task complexity and whole-body engagement. These effects are of high clinical value, as gait speed and balance duration are strong predictors of fall risk and mobility independence in LTCF populations.

Hand dexterity also showed meaningful enhancement, as evidenced by significant gains in the Box and Block test and a small group × time interaction effect. While modest in effect size, these changes carry functional importance. Tasks such as tug-of-war, rock-paper-scissors, and disc-catching required fine finger and wrist coordination under time pressure, thereby challenging and reinforcing rapid motor planning and visuomotor integration. These activities may have directly translated into improved hand control during activities of daily living, such as manipulating utensils or securing clothing fasteners.

The observed improvements in cognitive performance were not only statistically significant but also clinically interpretable when evaluated against established benchmarks. For the MoCA, participants demonstrated a mean score increase of over 2 points following the intervention—exceeding the commonly accepted minimally important difference of 2 points used in clinical research to denote perceptible improvement in cognitive status [51]. The gain is especially relevant in populations with MCI or early-stage dementia, where even modest changes may signal improvements in functional cognition, such as medication management, attention to safety cues, or basic financial planning. Similarly, participants exhibited a mean improvement of over 8 points on the CASI compared to baseline, corresponding to approximately 74% of the minimal detectable change (MDC) value of 11.6 points previously identified in dementia populations [38]. While the gain does not fully exceed the MDC threshold at the individual level, it indicates a robust group-level change, especially in light of the known variability in cognitive trajectories among institutionalized older adults.

Beyond improvements in overall cognitive scores, domain-specific patterns of change were also evident. Rather than uniform enhancement, the observed patterns suggest that the exergame-based intervention selectively engaged and strengthened a core set of cognitive processes—namely executive control, attentional regulation, memory updating, and visuospatial integration. The design of WarioWare: Move It! demands rapid response inhibition, visual tracking, rule-switching, and continuous adaptation to novel stimuli, all of which closely align with the targeted cognitive domains. Such gameplay simulates real-time cognitive-motor coordination and elicits engagement of distributed neural systems, including the frontoparietal control network, prefrontal regions associated with working memory, and posterior parietal areas supporting visuospatial mapping. Despite both MoCA and CASI captured improvements in shared domains (e.g., attention and memory), some divergence in the pattern of domain-specific gains also emerged, suggesting that the intervention may have differentially engaged complementary cognitive systems. MoCA reflected stronger improvements in visuospatial-executive functioning, attentional control, and delayed recall, which are associated with short-term adaptive flexibility and rapid working memory updating. In contrast, CASI revealed more pronounced gains in long-term memory, mental manipulation, and visual construction, suggesting a potential enhancement of semantic memory retrieval and structured cognitive planning. Far from being contradictory, these distinctions illustrate the multidimensional mechanism of the intervention, which engages multiple cognitive pathways through its integration of physical actions, multisensory stimuli, and structured variability. Taken together, these findings underscore the potential of embodied, gamified training to induce multifaceted cognitive gains, with implications for tailoring interventions to specific cognitive targets. In institutional care settings where cognitive stimulation is often constrained, such approaches offer a feasible and engaging alternative to traditional training, with the added advantage of addressing both mental and motor functions simultaneously.

Practical implications for geriatric rehabilitation in rural and resource-limited settings

The implementation of the intervention within rural LTCFs in China, which are frequently characterized by limited medical infrastructure, low therapist-to-resident ratios, and inadequate access to structured cognitive-motor training, highlights both the feasibility and the practical importance of culturally responsive rehabilitation approaches. Rural LTCFs commonly face sociocultural challenges, such as digital unfamiliarity among older adults, lower average literacy, and a strong reliance on informal caregiving. These factors typically restrict the applicability of traditional rehabilitation programs that require professional supervision, extensive equipment, or high levels of user autonomy. The high engagement and completion rates observed in this study thus suggest that the WarioWare: Move It! protocol was well-aligned with the preferences, capabilities, and care environments of this population.

Several features of the intervention likely contributed to its contextual success. First, its intuitive and playful interface, humorous content, and rapid microgame cycles reduced cognitive demand while fostering curiosity and enjoyment. It was especially valuable in a rural Chinese context where older adults may have limited prior exposure to digital games or formal therapy. Second, the seated gameplay mode, lack of wearable sensors, and reliance on gross motor gestures enabled wide participation without increasing fall risk or requiring specialized equipment. These design characteristics are critical in facilities with limited rehabilitation expertise and space. Third, the structured inclusion of cooperative and competitive gameplay within the intervention created meaningful opportunities for interpersonal interaction, which not only supported peer engagement but also helped alleviate the psychological isolation that is commonly observed among residents in rural long-term care settings, where social stimulation is often limited.

Importantly, the insights drawn from this study extend beyond its immediate setting. Anchored in low-cost commercial hardware, adaptable physical requirements, and socially engaging content, the intervention model exemplifies a scalable, game-based rehabilitation approach that is both effective and feasible in resource-limited settings. While certain contextual factors (e.g., cultural humor resonance, language interface) may require localization for broader dissemination, the core principles of adaptability, multisensory stimulation, and intrinsic motivation are likely transferable across diverse cultural and infrastructural settings.

Taken together, this study not only provides empirical support for the use of exergames in resource-constrained LTCFs, but also highlights the importance of aligning rehabilitation interventions with the sociocultural and infrastructural realities of target populations. Future research should explore cross-cultural adaptations of this approach, as well as long-term implementation strategies for sustainable integration into elder care practices globally.

Limitations and future work

Our study has several limitations that warrant consideration, along with areas for future improvement. First, although the total sample size met the requirements for statistical analyses, the distribution across cognitive subgroups was imbalanced, with a disproportionately larger number of participants in the MCI group compared to the mild and moderate dementia groups. The imbalance largely reflects the practical challenges associated with recruiting and retaining individuals with more advanced cognitive impairment in long-term care settings. Many residents with dementia were excluded due to unmet inclusion criteria, such as the inability to provide informed consent (independently or via proxy), comprehend task instructions, or participate safely in physical activity. Additional barriers included behavioral disturbances (e.g., apathy, disorientation, agitation) and functional limitations that impeded consistent engagement. While subgroup analyses were still conducted, the uneven distribution may have affected the statistical power to detect subtle differences between groups and should be considered when interpreting the results. Second, despite the exergames have positively affected specific physical and cognitive functions, they cannot fully replace traditional rehabilitation methods (e.g., physical or cognitive-behavioural therapy). Future intervention designs could consider integrating exergames with conventional rehabilitation techniques, creating a more diversified and comprehensive rehabilitation plan to meet the broader needs of older adults. Third, although all outcome assessments were conducted by trained evaluators blinded to group allocation, the same versions of cognitive assessment tools (e.g., MoCA, CASI) were used at each time point, which may have introduced practice effects, particularly in participants with higher cognitive reserve. Future studies should consider using alternate or parallel forms of cognitive tests, or incorporating objective cognitive or neurophysiological measures, to improve the validity and interpretability of cognitive outcomes. Finally, the social interaction features of “WarioWare: Move It!” (e.g., competition, cooperation, and rivalry) enhance rehabilitation outcomes. However, preferences for competition or collaborative modes may vary among older adults, particularly those who are more introverted or have lower social engagement, making these modes less appealing to some. Future game iterations could incorporate personalized social interaction options to ensure each participant receives an optimal experience, boosting engagement.

Conclusions

The “WarioWare: Move It!” training program significantly improves physical flexibility, joint range of motion, motor coordination, hand dexterity, and cognitive function among older adults in rural LTCFs. The intervention offers a novel approach to providing supportive health care in resource-limited settings. Specifically, for patients with mild cognitive impairment or dementia, the intervention not only aids in delaying functional decline but also enhances cognitive function and quality of daily living through multisensory stimulation and dynamic interaction. Furthermore, the incorporation of exergames marks an innovative shift in elderly care models, offering a novel, low-demand intervention method that effectively addresses the shortcomings of traditional care approaches. These findings support the integration of exergames into routine care practices and provide a viable, technology-driven solution to address the challenges of aging health in institutional care settings worldwide.

Abbreviations

LTCFs

Long-term care facilities

MCI

Mild cognitive impairment

SD

Standard deviation

ANOVA

Analysis of variance

MoCA

Montreal cognitive assessment

CASI

Cognitive abilities screening instrument

GDS

Global deterioration scale

Authors’ contributions

A.L.: Conceptualization, Investigation, Writing - Original Draft, Supervision; W.Q.: Resources, Writing - Original Draft; J.L.: Resources, Investigation, Writing - Original Draft, Project administration; Y.G.: Methodology, Data Curation, Writing - Review & Editing; Y.Q.: Formal analysis, Writing - Review & Editing. J.Z.: Funding acquisition, Formal analysis, Writing - Review & Editing. All authors reviewed and approved the manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Data availability

The data used in the study is not publicly available, but the data used and/ or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Biomedical Ethics Review Committee of Taiyuan University of Technology (Approval No. 20230641). Given the inclusion of older adults with suspected or diagnosed cognitive impairment, informed consent was obtained for all participants prior to their enrollment in the study. Individuals with sufficient cognitive capacity provided written informed consent independently. For participants with impaired decision-making ability, written consent was obtained from their legally authorized representatives (e.g., family members or appointed guardians), in accordance with institutional review board guidelines. The consent procedure was conducted by trained research staff in a quiet and private setting.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.