The current systematic review was registered on PROSPERO with the following registration (CRD42023478330 [47]).

The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines were followed for this review. The PRISMA item checklist can be found in Checklist 1. The search time frame was from inception to November 1, 2023. The goal was to identify recent studies on the effects of VR training for enhanced balance, strength, and mobility at home among older adults with frailty and prefrailty. Two authors independently performed searches in the following databases: Scopus, Web of Science, and PubMed.

The search terms were specific to each database. The following is an example of the search terms used in Scopus: risk of fall OR balance OR strength OR function AND frail OR prefrail AND older adult AND virtual reality OR video games OR mobile game. A detailed overview of the search terms and strategies used is provided in Multimedia Appendix 1.

The study comprised all English-language papers, including those that used a single-group design in which a VR as a home-based exercise intervention was compared with no intervention or other interventions for enhanced strength, balance, and mobility among older adults with frailty or prefrailty. The PICOS (Population, Intervention, Comparison, Outcome, and Study Design) framework for the current review was as follows:

Table 1 summarizes the inclusion and exclusion criteria structured according to the PICOS framework.

The included studies comprised male or female older adults with a mean age of 65 years or older, described as older adults with frailty, older adults with prefrailty, aged, geriatric, or older adults living in the community, independently, in retirement centers or nursing homes. Studies that included participants with specific medical conditions, such as stroke, Parkinson disease, or cognitive impairment, were excluded. Table 2 summarizes the frailty status of participants across the included studies.

The included studies explored various interventions to improve balance, mobility, and physical function; these included video-guided exercises with resistance bands, VR balance training guided by a physiotherapist, or sensor-based exergames exercised at home. Each intervention was compared to traditional exercise programs or participants’ usual activities to assess their effectiveness.

The included studies assessed the effectiveness of various VR interventions for older adults by measuring changes in physical function and mobility. Outcomes such as the knee extension strength test and Sit-to-Stand test were used to assess lower extremity strength, the Berg Balance Scale (BBS) test was used to assess balance, and the Timed Up and Go (TUG) test was used to assess functional mobility.

The Physiotherapy Evidence Database (PEDro) assessment tool was used to assess the methodological quality of the included studies [54]. The total PEDro score reflected the quality of the study as follows: a total score of ≥6 indicated high quality, 4‐5 represented fair quality, and ≤3 indicated poor quality [55].

The reviewers independently assessed the trials for eligibility by reviewing the titles and abstracts. If a paper title or abstract was deemed relevant, the full text was retrieved for evaluation against the inclusion and exclusion criteria. Any disagreement between the authors was resolved by the lead author. A data extraction form was created, and the data were extracted by the independent reviewers.

A random-effects meta-analysis was performed using Review Manager (version 5.4; Cochrane) software. Three primary outcomes were included: BBS, TUG, and Chair Stand (CS) tests. The purpose was to identify the mean difference (MD) in balance, risk of falls, and strength between VR groups and conventional intervention or control groups, and also to determine the overall treatment effect size. Heterogeneity was assessed with the I² index, which has 4 classification levels: unimportant heterogeneity (0%‐40%), moderate heterogeneity (30%‐60%), substantial heterogeneity (50%‐90%), and considerable heterogeneity (75%‐100%) [56]. Effect sizes were calculated for all studies using Cohen d, with <0.2 indicating a “trivial” effect size, with 0.2 indicating a “small” effect size, 0.5 indicating a “medium” effect size, and 0.8 indicating a “large” effect size [57].

A total of 1063 papers were identified as relevant; 40 were duplicates. The remaining 1023 papers were screened. After the initial screening, 1017 papers were excluded based on the titles and abstracts. The final review included 6 papers. The selection process for this systematic review is presented in the flow diagram in Figure 1.

The mean PEDro score was 5.6 (SD 1.3) with 5 studies [48,50-53] graded as high quality and 1 [49] as poor quality. Table 3 presents the results of the quality assessment of the included studies.

This current systematic review aimed to identify studies that have investigated the use of VR training at home to improve balance, strength, and mobility among older adults with frailty and prefrailty. Although few studies have focused on this aim, the findings showed that VR was effective at improving balance but not strength and mobility. One potential reason for this could be due to limited experience with independent technology use among older adults and therapists, a challenge that has been validated in previous studies [41,58]. Nevertheless, the studies included in this review are examples of how technological advancements can reshape health care delivery and demonstrate that certain VR interventions can be used safely with older adults, as no study reported any safety hazards.

The findings from this review showed that home-based VR intervention produces great variability in effectiveness on different outcomes. On balance, both [51,52] found a significant effect on balance in the VR group when using the BBS. This finding aligns with a recent review [59], which also found a significant improvement in BBS scores following VR intervention in older adults with balance impairments. Similarly, another study [60] found that VR intervention significantly improved BBS scores in older adults residing in nursing homes compared with those living in communities. The populations studied in these studies were similar, indicating that these findings may be generalizable.

For strength outcomes, the CS test did not show significant improvements in 4 of the included studies. In the study by Vestergaard et al [48], the lack of improvement may be due to insufficient progression in the exercises included in the VR training group. Without adequate progression or intensity, participants may not have experienced meaningful gains in strength or functional mobility. However, this remains a possible explanation rather than a definitive conclusion. In the study by Gschwind et al [50], although the Step Mat Training (SMT) group demonstrated significant improvements in sit-to-stand times compared to the control group, no significant differences were found between the 2 intervention groups or between the intervention group and the control group. This limited improvement could suggest that Kinect training may not have been as effective in enhancing strength and balance. The intensity and specificity of exercises delivered through VR interventions are likely contributing factors. While some studies may have used tailored and progressively challenging tasks to enhance training efficacy, others may have implemented more generic routines that did not adequately target the specific mobility deficits of participants.

In contrast to mobility outcomes, home-based VR training’s effect on the TUG showed variation in the included studies. The lack of significant improvement could be attributed to the small sample size [49,50], which limited the ability to detect changes. This variation in effect on TUG could also be due to the specific design and characteristics of the VR interventions implemented in these studies. For instance, 1 study [49] used short-duration interventions of 10‐20 minutes per session, while another [51] used longer-duration interventions of 30 minutes per session. Similarly, the VR effect on knee extension strength did not show significant improvements in the 2 studies [49,50]. In the study by Schoene et al [49], the intervention did not specifically target knee extension strength or range of motion, focusing more on stepping performance and cognitive parameters related to fall risk. In contrast, Gschwind et al [50] found no significant differences between the 2 intervention groups or between the SMT group and the control group. This could be due to the SMT program focusing on proprioception, cognitive processing, and balance, which may have shifted the emphasis away from targeting strength. In contrast, the Kinect program may have placed more emphasis on strength-building exercises [50].

The current meta-analysis clearly highlights 2 aspects: the frail and prefrail status of participants and the home-based delivery of VR training. The importance of targeting older adults with frailty or prefrailty originates from the increased vulnerability among this demographic to functional decline and falls, emphasizing the need for tailored rehabilitation interventions [61,62]. Previous systematic reviews and trials examining VR interventions in rehabilitation centers among older populations, rather than specifically individuals with frailty or prefrailty, have reported varied findings. For instance, a recent systematic review [59] showed significant improvements in balance, strength, and mobility among older adults participating in center-based VR programs. Similarly, an RCT [60] delivered in nursing homes showed improvements in balance outcomes, specifically on the BBS, following structured VR interventions. However, such site-specific studies often comprise higher levels of supervision and formal instruction from a therapist, which are not necessarily available in home-based settings. This key distinction highlights the importance of evaluating home-based VR effectiveness independently. Notably, this review found challenges in standardizing intervention intensity and ensuring adherence and safety without direct professional oversight. This is in line with recent findings [63], which reported technology use challenges among older adults without supervision, potentially limiting the effectiveness of home-based VR interventions. Furthermore, studies conducted among older adults often report increased baseline functional ability compared to populations with frailty and prefrailty, possibly overestimating intervention effectiveness. For instance, the study by Donath et al. [64] reported significant improvements in functional performance following a supervised VR training program in healthy older adults. In contrast, our review’s targeted populations with frailty and prefrailty showed smaller effect sizes in mobility outcomes (TUG), possibly due to lower initial functional status.

Overall, our study revealed that VR exercises significantly enhanced balance among older adults compared to those who engaged in regular exercises or remained inactive. However, there were no clear and significant differences in strength and functional mobility. This suggests that while VR training holds potential, its design and implementation require careful consideration to ensure progressive difficulty and proper evaluation through controlled studies.

First, only studies published in English were included, raising the possibility of language bias. This might have caused the exclusion of related studies. While this decision was made to ensure precise analysis of findings, future reviews could compensate for this limitation by using translation resources or multilingual reviewers.

Second, there was considerable heterogeneity in intervention design across the included studies. Variations in VR training, session duration, frequency, and outcome measures made it difficult to directly compare results. The heterogeneity in study protocols could have been a factor in the inconsistent findings. Future studies are advised to design standardized intervention protocols to mitigate these differences.

Third, the small sample sizes in multiple studies limited the statistical power to identify significant outcomes. While we used rigorous quality assessment (eg, PEDro scale) to identify high-quality studies, the small sample size may have resulted in underpowered analyses. Future studies should include larger, adequately powered samples to strengthen the validity of the outcomes.

Fourth, this review excluded gray literature and unpublished studies, which may have introduced publication bias. Although we conducted a comprehensive search across multiple databases to mitigate this risk, future reviews should consider searching clinical trial registries and preprint servers to identify ongoing or unpublished work.

We should aim to standardize VR protocols. Future studies should develop structured VR interventions with gradual difficulty and standardized outcomes to enhance consistency and comparability across trials.

We should aim to assess long-term effects. Future studies should assess adherence, sustainability of outcomes, and cost-effectiveness over extended periods.

We should aim to enhance engagement and adherence. VR’s interactive nature offers potential benefits beyond physical improvement, including motivation and adherence.

We should aim to explore clinical adoption. Future research should investigate health care providers’ perceptions, barriers, and facilitators to implementing VR-based training in routine geriatric care.

This systematic review aimed to evaluate the effectiveness of home-based VR training in improving balance, strength, and mobility among older adults with frailty and prefrailty. The findings suggest that VR interventions are consistently effective in improving balance but show limited evidence for improving strength and mobility. These outcomes were influenced by variability in intervention design, duration, and intensity across studies.