This paper is in the following e-collection/theme issue:

Digital Health Reviews (1444) Extended Reality, Virtual Reality and Virtual Worlds (621) Stroke (182) Emerging Technologies for Rehabilitation (292) Innovations in Exercise Rehabilitation (228)

Published on in Vol 27 (2025)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/66802, first published .
Virtual Reality Enhanced Exercise Training in Upper Limb Function of Patients With Stroke: Meta-Analytic Study

Virtual Reality Enhanced Exercise Training in Upper Limb Function of Patients With Stroke: Meta-Analytic Study

Virtual Reality Enhanced Exercise Training in Upper Limb Function of Patients With Stroke: Meta-Analytic Study

Authors of this article:

Shiqi Xu1, 2 Author Orcid Image ;   Yanwen Xu1 Author Orcid Image ;   Ruyi Wen3 Author Orcid Image ;   Jun Wang1 Author Orcid Image ;   Yuyu Qiu2 Author Orcid Image ;   Chetwyn CH Chan4 Author Orcid Image
Article Authors Cited by Tweetations Metrics

    Review

    1Department of Rehabilitation Medicine, Wuxi Ninth People’s Hospital Affiliated to Soochow University, Wuxi, Jiangsu, China

    2Wuxi School of Medicine, Jiangnan University, Wuxi, Jiangsu, China

    3School of Rehabilitation Medicine, Nanjing Medical University, Nanjing, China

    4Department of Psychology, The Education University of Hong Kong, Hong Kong SAR, China

    *these authors contributed equally

    Corresponding Author:

    Chetwyn CH Chan

    Department of Psychology

    The Education University of Hong Kong

    Lo Ping Road, Tai Po, New Territories

    Hong Kong SAR

    China

    Phone: 852 2948 6322

    Email: cchchan@eduhk.hk


    Background: Recovery of upper limb function after stroke secondary to ischemia or hemorrhage is crucial for patients’ independence in daily living and quality of life. Virtual reality (VR) is a promising computer-based technology designed to enhance the effects of rehabilitation; however, the results of VR-based interventions remain equivocal.

    Objective: This study aims to review the plausible factors that may have influenced VR’s therapeutic effects on improving upper limb function in patients with stroke, with the goal of synthesizing an optimal VR intervention protocol.

    Methods: The databases PubMed, EMBASE, Web of Science, and Cochrane Library were queried for English-language papers published from May 2022 onward. Two reviewers independently extracted data from the included papers, and discrepancies in their findings were resolved through consensus during joint meetings. The risk of bias was assessed using the Physiotherapy Evidence Database Scale and the Methodological Index for Non-Randomized Studies. Outcome variables included the Action Research Arm Test, Box-Block Test, Functional Independence Measure, Upper Extremity Fugl-Meyer Assessment, and Wolf Motor Function Test. The plausible factors examined were age, total dosage (hours), trial length (weeks), session duration (hours/session), frequency (sessions/week), and VR content design. The Bonferroni adjustment was applied to P values to prevent data from being incorrectly deemed statistically significant.

    Results: The final sample included 15 articles with a total of 1243 participants (age range 48.6-75.59 years). Participants in the VR therapy (VRT) group (n=455) demonstrated significantly greater improvements in upper limb function and independence in activities of daily living compared with those in the conventional therapy group (n=301). Significant factors contributing to improved outcomes in upper limb function were younger age (mean difference [MD] 5.34, 95% CI 2.18-8.5, P<.001; I2=0%), interventions lasting more than 15 hours (MD 9.67, 95% CI 4.19-15.15, P<.001; I2=0%), trial lengths exceeding 4 weeks (MD 4.02, 95% CI 1.39-6.65, P=.003; I2=15%), and more than 4 sessions per week (MD 3.48, 95% CI 0.87-6.09, P=.009; I2=0%). However, the design of the VR content, including factors such as the number of features (eg, offering exercise and functional tasks; individualized goals; activity quantification; consideration of comorbidities and baseline activity level; addressing patient needs; aligning with patient background such as education level; patient-directed goals and interests; goal setting; progressive difficulty levels; and promoting self-efficacy), did not demonstrate significant effects (MD 3.89, 95% CI –6.40 to 1.09; effect Z=1.36, P=.16).

    Conclusions: Greater VR effects on improving upper limb function in patients with stroke were associated with higher training doses (exceeding 15 hours) delivered over 4-6 weeks, with shorter sessions (approximately 1 hour) scheduled 4 or more times per week. Additionally, younger patients appeared to benefit more from the VR protocol compared with older patients.

    J Med Internet Res 2025;27:e66802

    doi:10.2196/66802

    Keywords

    virtual reality; stroke; upper limb function; exercise training; meta-analysis 


    Recovery of upper limb function after stroke is crucial for patients’ independent daily living and quality of life [1-3]. Previous studies have demonstrated the positive effects of incorporating neurodevelopmental, proprioceptive neuromuscular, and motor relearning theories in designing interventions to improve patients’ upper limb function [4-7]. Advances in technology have enabled virtual reality (VR) to serve as a common medium for delivering interventions aimed at enhancing upper limb function [8]. VR is a promising computer technology that allows users to interact with a simulated multisensory environment [9]. It is designed to enhance the effectiveness of rehabilitation and provide feedback on performance. However, a review of clinical studies examining the use of VR as an intervention for patients with stroke produced inconsistent findings. Compared with usual practice, VR interventions were found to be more effective in improving upper-extremity and hand functions in patients with stroke [10-12]. By contrast, 1 study reported that VR interventions were effective in improving cognitive functions but not motor functions in patients with traumatic brain injury [13], while another study found no significant differences between VR and conventional therapy for patients with stroke [14].

    A review of recent studies suggested that the inconsistent results described above may stem from variations in the design, content, duration, and intensity of VR interventions [14], as well as patient-specific characteristics such as independence level and emotional state [12]. For example, VR exercise training was found to be more effective than usual practice for improving upper limb function in patients with stroke only when they participated in a high-dose program (eg, 15 hours or more) [15], whereas this effect was not observed with a low-dose program (eg, 12 hours) [16]. In the high-dose study, the VR exercise training included a variety of limb movements with varying levels of difficulty. Conversely, the low-dose study used a VR protocol consisting of 3 modes of hand and arm movements performed at different speeds. Inconsistent findings are also evident in studies evaluating the effects of VR in patients with Parkinson’s disease. For example, a VR game using a 1-group pre- and posttest design demonstrated improvements in hand grip strength, dexterity, and speed among patients with stroke [17]. The VR game comprised 4 activities delivered in sequence: the reach game, the sequence game, the grab game, and the flip game. The intervention involved 30-minute individual sessions conducted 3 times per week over 6 weeks (18 sessions, totaling 9 hours). These results contrast with those of a randomized controlled trial (RCT), which found no significant effects [18]. The RCT protocol involved tracking and touching moving objects displayed on a screen while maintaining a standing balance. The schedule consisted of 40-minute individual sessions, 3 sessions per week over 6 weeks (18 sessions, totaling 12 hours). Many national clinical guidelines for stroke emphasize the importance of applying recovery principles. Beyond dosage (ie, duration per session, frequency, and overall intervention period), the design of VR training content appears to influence treatment outcomes. For instance, the 2023 edition of the National Clinical Guideline for Stroke for the United Kingdom and Ireland [19] recommends therapy for motor recovery and function that incorporates individualized content, a variety of practice modes, intensity, and the practice of functional skills. Additionally, studies have highlighted the critical role of repetitions in reshaping neuronal structures and enhancing motor system function following brain injury [20-22]. Research on nonprimates has shown that performing repeated reaching tasks can induce synaptogenesis and changes in cortical representations [23,24]. In human studies, evidence suggests that functional recovery of the upper limbs, driven by the brain’s plasticity, requires at least 300 repetitions per day [25] or 100 active movements or more per treatment session [26].

    The inconsistent findings regarding the effects of VR-based interventions motivated us to undertake this meta-analytic study. We aimed to examine 6 plausible factors that might influence the therapeutic effects of VR on improving upper limb function in patients with stroke. These factors included patients’ age, dosage (hours), delivery schedule (session duration, frequency, and trial length), total dosage (hours), and content design. We hypothesized that all these factors would significantly impact intervention outcomes. Identifying the factors contributing to variability in intervention outcomes would enhance the content design of VR interventions and provide guidance on setting appropriate dosages and schedules to achieve better treatment results for patients.


    Design

    This study followed the scope and methods outlined in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement and its accompanying checklist [27].

    Search Strategy

    The relevant materials for review and analysis were extracted on May 7, 2022, from PubMed (1966 to present), EMBASE via Ovid (1974 to present), Web of Science (1956 to present), and the Cochrane Library databases (no date restriction). The medical terms and free-text search terms were related to VR, upper limb, and rehabilitation. Detailed search strategies for each database are provided in Multimedia Appendix 1.

    Inclusion and Exclusion Criteria

    The inclusion and exclusion criteria set for the studies are presented in Textboxes 1 and 2.

    Textbox 1. Inclusion criteria.
    • Randomized controlled trial and group comparison studies.
    • Results published in English in peer-reviewed journals.
    • The treatment group used virtual reality–enhanced exercise alone or combined with conventional therapy, while the control group used conventional therapy alone or without treatment.
    • Patients showed upper limb functional disabilities with outcome variables including the Action Research Arm Test, Box-Block Test, Functional Independence Measure, Upper Extremity Fugl-Meyer Assessment, and Wolf Motor Function Test.
    • Means, SD, and effect sizes were presented; or means, SEs, t values, or P values were presented; or range or 95% CIs were presented.
    Textbox 2. Exclusion criteria.
    • Abstracts, case reports, or review studies.
    • Non-English or nonbilingual journal articles.
    • Research design without group comparison.
    • Patients without stroke or measuring lower limb functions of patients with stroke.
    • Missing data or additional information (means, SD, and effect sizes) cannot be obtained from the corresponding author within 1 month after initial contact.

    Study Selection

    All studies were imported into the EndNote referencing software (Thomson Research Soft, version X9), and duplicates were identified and removed. Two reviewers, SX and JW, co-authors of this paper, independently conducted the relevance screening. The screening process began with an evaluation of the title and abstract content of each publication based on the inclusion and exclusion criteria outlined above. Reviewers were instructed to read the full text if the abstract contained ambiguous content. Discrepancies between the 2 reviewers were resolved through consensus during joint meetings.

    Data Extraction

    Two research team members (SX and YX) independently extracted data from the included papers. For each paper, the researchers identified the trials and specific content, recording them in preformatted tables. The recorded information included trial design and setup, sample sizes of all groups, intervention program details, overall characteristics, and clinical outcome results. Discrepancies between the 2 reviewers were resolved through consensus during joint meetings.

    Assessment of Risk of Bias

    Two research team members (YX and YQ) independently assessed the risk of bias in the included trials using the Physiotherapy Evidence Database (PEDro) Scale [28] and the Methodological Index for Non-Randomized Studies (MINORS) [29]. The 11-item PEDro Scale evaluates the quality of the trials, with a score of “1” indicating the criterion is met and “0” indicating it is not. PEDro quality grades are categorized as high (6-8: good; 9-10: excellent), fair (4-5: acceptable), and poor (≤4). For nonrandomized trials, the first 8 items of the MINORS were used to assess methodological quality (risk of bias). Each MINORS item is scored as follows: “0” for not reported, “1” for reported but insufficient, and “2” for reported and sufficient. For a single-group trial, the maximum score is 16, while for a 2-group trial, it is 24. A score of 0-9 for a single-group trial indicates poor quality and 10-16 indicates high quality. In this study, the cut-off for inclusion was set at high quality or above: a score of 6 or higher on the PEDro Scale or 10 or higher on the MINORS [30]. Discrepancies in the scores assigned by the 2 reviewers were resolved through consensus during joint meetings.

    The principles of therapy content (motor recovery and function) outlined in the 2023 edition of the National Clinical Guideline for Stroke for the United Kingdom and Ireland [19] were adopted as feature criteria to guide the analysis of VR content. Ten content features were identified: offering exercise and functional tasks; setting individualized goals; quantifying activity; considering comorbidities and baseline activity levels; incorporating patient needs; matching the patient’s background, such as education level; aligning with the patient’s goals and interests; involving goal setting; including progressive difficulty levels; and promoting self-efficacy. The content features of each study were collated and classified into 3 categories: “+” for few features (2-3), “++” for medium features (4-5), and “+++” for many features (≥6).

    Data Synthesis and Analysis

    We conducted analyses using Review Manager (RevMan) version 5.3 software. As some studies [14,31-34] reported results in terms of the median, minimum, maximum, or IQR, we converted these to means and SDs as proposed in the study by Wan et al [35]. For studies reporting only means, SEs, t values, or P values, we computed the SDs [31,32,35]. For studies reporting only ranges or 95% CIs, we calculated the means [31,32,35]. Statistical heterogeneity for each meta-analysis was assessed using the Cochran Q test and the I2 statistic. Fixed-effect models were applied when heterogeneity among trials was nonsignificant (I2<50%), while random-effect models were used for significant heterogeneity (I2>50%). The outcome variables included the Action Research Arm Test (ARAT), Box and Block Test (BBT), Functional Independence Measure (FIM), Upper Extremity Fugl-Meyer Assessment (FMA-UE), and Wolf Motor Function Test (WMFT). All results were reported as mean differences (MDs) with 95% CIs. The plausible factors considered were age (with the World Health Organization defining older age as 60 years and above), total dosage (hours), trial length (weeks), duration of a session (hours/session), frequency (sessions/week), and VR content design. The cut-off value was set based on the findings of previous literature [36-39] and the principle that at least two articles must be included per subgroup. As only 15 studies were available for analysis, multiple comparisons were conducted separately for each of the 6 factors to meet this criterion. Statistical significance for the meta-analysis (whole group) was set at P<.05. For the subgroup analyses, the potential proliferation of type I and type II errors across the 6 comparisons was controlled by applying the Bonferroni adjustment, setting statistical significance at P<.008 (0.05 divided by 6).


    Study Selection

    Details of the steps taken to select the studies are presented in the PRISMA flowchart (Figure 1; also see Multimedia Appendix 2). A total of 745 studies were retrieved from 4 databases through an electronic search. Of these, 237 studies were excluded due to duplication. An additional 303 studies were excluded after reviewing their titles and abstract contents. Full-text screening of the remaining 205 studies resulted in the exclusion of 185 studies for not meeting the inclusion criteria. Among the remaining 20 studies, 5 were further excluded for being rated below the “good” cut-offs on the PEDro and MINORS scales. The final number of studies included in the analyses was 15 [11,16,33,34,40-50], consisting of 12 RCTs and 3 non-RCTs.

    ‎
    Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart showing 15 records entering the analyses. VR: virtual reality.

    Characteristics of the Included Trials

    The 15 studies included a total of 28 groups, comprising 15 VR therapy (VRT) groups and 13 conventional therapy (CON) groups, with 1203 participants in total (Table 1). All studies were published between 2013 and 2022. At the total group level (14 studies), the VRT and CON groups included 411/278 and 294/177 males/females, respectively. One study did not report gender as a grouping variable and was therefore excluded from the gender analysis at the group level. The average ages of participants in the VRT groups ranged from 49.0 to 73.0 years, while those in the CON groups ranged from 53.4 to 75.6 years. The intervention durations for the VRT groups varied from 2 weeks to 3 months. Fourteen studies [11,16,33,34,40-47,49,50] involved patients with stroke secondary to ischemia or hemorrhage, while 1 study [48] included only patients with stroke secondary to ischemia.

    Table 1. Characteristics of the trials contained in 15 studies satisfying the inclusion criteria.
    StudyStudy designParticipantsAge (years), meanInterventionControlDosageTotal dosage (hours)Outcomes included in this review
    Intervention/control (no/no)Intervention/control (yes/yes)
    Brunner et al [40]RCTa62/5862/62YouGrabber systemConventional training60 minutes/day, 4 days/week, 4 weeks16 ARATb, BBTc, and FIMd
    Abd El-Kafy et al [41]RCT20/2054.13/53.42 Conventional physiotherapy and training with robot-mediated VRe gamingConventional physiotherapy2 hours/day, 3 days/week, 12 weeks72 ARAT and WMFTf
    Gueye et al [42]RCT25/2566.56/68.12Armeo SpringConventional physiotherapy45 minutes/day, 4 days/week, 3 weeks9 FMA-UEg and FIM
    Hsu et al [11]RCT18/1752.9/56.9VR-based mirror therapy and usual careConventional occupational therapy and usual care50 minutes/day, 2 days/week, 9 weeks15FMA-UE and BBT
    Kiper et al [43]RCT23/2164.3/64.3VR rehabilitation system and traditional rehabilitationTraditional rehabilitation2 hours/day, 5 days/week, 4 weeks40FMA-UE and FIM
    Kiper et al [44]RCT68/6862.5/66VR rehabilitation system and traditional rehabilitationTraditional rehabilitation2 hours/day, 5 days/week, 4 weeks40FMA-UE and FIM
    Rong et al [45]RCT20/2056.25/62.3Camera-based mirror visual feedback and robot-assisted trainingSham-mirror visual feedback and robot-assisted training1.5 hours/day, 5 days/week, 4 weeks30FMA-UE and FIM
    Schuster-Amft et al [16]RCT22/3261.3/61.2VR-based training systemConventional therapy45 minutes/day, 4 days/week, 4 weeks12BBT
    Sin and Lee [46]RCT18/1771.78/75.59Xbox Kinect and conventional occupational therapyConventional occupational therapy1 hour/day, 3 days/week, 6 weeks18FMA-UE and BBT
    Taveggia et al [47]RCT27/2773/68Armeo spring and conventional treatmentConventional treatment1 hour/day, 5 days/week, 6 weeks30FIM
    Yao et al [48]RCT20/2063/66.2C-tDCSh and VR therapy and conventional occupational and physical therapiesSham tDCS and VR and conventional occupational and physical therapies20 minutes/day, 5 days/week, 2 weeks3.33ARAT and FMA-UE
    Zheng et al [49]RCT58/5465.4/66.2Low-frequency repetitive transcranial magnetic stimulation and VR and standard rehabilitation therapy for strokeSham repetitive transcranial magnetic stimulation and VR training and standard rehabilitation therapy for stroke2 hours/day, 6 days/week, 4 weeks48WMFT
    Turolla et al [50]Non-RCT263/11360.2/65.4Upper limb conventional and reinforced feedback in the virtual environment therapiesUpper limb conventional therapy2 hours/day, 5 days/week, 4 weeks40FMA-UE and FIM
    Borstad et al [33]Non-RCT16/N/Ai49/N/AConstraint-induced movement therapy (recovery rapids and highly trained practitioners)N/A3 hours/day, 5 days/week, 2 weeks30ARAT and WMFT
    Sebastián-Romagosa et al [34]Non-RCT51/N/A60.52/N/ABrain-computer interface systemsN/A1 hour/day, 2 days/week, 3 months25FMA-UE

    aRCT: randomized controlled trial.

    bARAT: Action Research Arm Test.

    cBBT: Box-Block Test.

    dFIM: Functional Independence Measure.

    eVR: virtual reality.

    fWMFT: Wolf Motor Function Test.

    gFMA-UE: Upper Extremity Fugl-Meyer Assessment.

    hC-tDCS: cathodal transcranial direct current stimulation.

    iN/A: not applicable.

    Risk of Bias

    The mean PEDro Scale score for the 12 studies with a randomized clinical trial design was 7.2 (Table 2). All these studies reported the method of randomization, baseline comparability, blinding of outcome assessors, participant dropout rates of less than 15%, between-group outcome analysis, and point estimates with variability. Most of these studies also reported adequate assignment concealment (n=7, 58%) and intention-to-treat analyses (n=4, 33%). Among the studies with a nonrandomized clinical trial design (n=3), it is noteworthy that they demonstrated clear research objectives and prospective data collection but lacked a follow-up period and did not perform sample size calculations (Table 3).

    Table 2. Physiotherapy Evidence Database Scale score for randomized clinical trials included in the review.
    StudiesTotal scoreMethodological qualityNumber of items of the Physiotherapy Evidence Database Scale
    1a2b3c4d5e6f7g8h9i10j11k
    Ikbali Afsar et al [14]5FairN/Al✓mN/A✓N/AN/A✓N/AN/A✓✓
    Alves et al [10]5FairN/A✓N/A✓N/AN/A✓N/AN/A✓✓
    Brunner et al [40]7GoodN/A✓✓✓✓N/AN/A✓N/A✓✓
    Abd El-Kafy et al [41]6GoodN/A✓✓✓N/AN/A✓✓N/A✓✓
    Ersoy and Iyigun [51]4PoorN/A✓✓✓N/AN/AN/AN/AN/A✓✓
    Gueye et al [42]6GoodN/A✓✓✓N/AN/AN/A✓✓✓✓
    Hsu et al [11]7GoodN/A✓✓✓N/AN/A✓✓N/A✓✓
    Junior et al [52]5FairN/A✓N/A✓N/AN/A✓N/AN/AXX
    Kiper et al [43]6GoodN/A✓N/A✓N/AN/A✓✓N/A✓✓
    Kiper et al [44]8GoodN/A✓✓✓N/AN/A✓✓✓✓✓
    Rong et al [45]8GoodN/A✓✓✓N/AN/A✓✓✓✓✓
    Schuster-Amft et al [16]8GoodN/A✓✓✓N/A✓✓✓N/A✓✓
    Sin and Lee [46]7GoodN/A✓N/A✓N/A✓✓✓N/A✓✓
    Taveggia et al [47]7GoodN/A✓✓✓✓N/AN/A✓N/A✓✓
    Yao et al [48]8GoodN/A✓✓✓N/A✓✓✓N/A✓✓
    Zheng et al [49]8GoodN/A✓N/A✓N/A✓✓✓✓✓✓

    aItem 1: Specified eligibility criteria.

    bItem 2: Random allocation.

    cItem 3: Concealed allocation.

    dItem 4:Baseline comparability.

    eItem 5: Participants were blinded.

    fItem 6: Therapists were blinded.

    gItem 7: Assessors were blinded.

    hItem 8: Adequate follow-up.

    iItem 9: Intention-to-treat analysis.

    jItem 10: Between-group comparisons.

    kItem 11: Point estimates and variability.

    lN/A: not applicable.

    mThe ‘✓’ symbol indicates that the item where it is found has been punctuated.

    Table 3. Methodological Index for Non-Randomized Studies Scale score for nonrandomized clinical trials included in the review.
    StudiesTotal scoreMethodological qualityNumber of items of the Methodological Index for Non-Randomized Studies Scalea
    1b2c3d4e5f6g7h8i9j10k11l12m
    Turolla et al [50]15/24Good222220002201
    Borstad et al [33]10/16Good22221100N/AN/AN/AN/A
    Kizony et al [53]7/16Poor21220000N/AN/AN/AN/A
    Sebastián-Romagosa et al [34]10/16Good22221100N/AN/AN/AN/A

    aThe items are scored 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate). The global ideal score is 16 for noncomparative studies and 24 for comparative studies.

    bItem 1: A clearly stated aim.

    cItem 2: Inclusion of consecutive patients.

    dItem 3: Prospective collection of data.

    eItem 4: Endpoints appropriate to the aim of the study.

    fItem 5: Unbiased assessment of the study endpoint.

    gItem 6: Follow-up period appropriate to the aim of the study.

    hItem 7: Loss to follow-up less than 5%.

    iItem 8: Prospective calculation of the study size.

    jItem 9: An adequate control group.

    kItem 10: Contemporary groups.

    lItem 11: Baseline equivalence of groups.

    mItem 12: Adequate statistical analyses.

    Meta-Analysis

    Whole-Group: Upper Limb Function (FMA-UE)

    Participants in the VRT group (n=455) demonstrated significantly greater improvement than those in the CON group (n=301) in FMA-UE scores (MD 3.80, 95% CI 1.47-6.13, P=.001; I2=7%; Figure 2; see also [11,16,40-50]). Within the VRT group, participants showed significant increases in FMA-UE scores following the intervention (MD 9.95, 95% CI 6.49-13.41, P<.001; I2=56%; Figure 3; see also [11,16,33,34,40-50]).

    ‎
    Figure 2. Forest plot of comparisons of between-group outcomes on upper limb function. CON: conventional therapy; VRT: virtual reality therapy.
    ‎
    Figure 3. Forest plot of comparisons of within-group outcomes on upper limb function.
    Whole Group: Upper Limb Activity Limitation (ARAT, FIM, WMFT, and BBT)

    Participants in the VRT group showed significantly greater improvements than those in the CON group in upper limb function measured by the ARAT (MD 3.68, 95% CI 1.87-5.48, P<.001; I2=0%; n [VRT]=102, n [CON]=98; Figure 2), independence in activities of daily living measured by the FIM (MD 4.86, 95% CI 2.92-6.81, P<.001; I2=46%; n [VRT]=488, n [CON]=332; Figure 2), and upper limb functional capacity measured by the WMFT (MD 6.16, 95% CI 4.03-8.29, P<.001; I2=0%; n [VRT]=78, n [CON]=74; Figure 2). However, there were no significant differences between the groups in upper limb finger dexterity measured by the BBT (MD 3.26, 95% CI –1.16 to 7.68, P=.15; I2=0%; n [VRT]=120, n [CON]=124; Figure 2).

    Among the VRT groups, participants demonstrated significant pre- and postintervention improvements in scores on the ARAT (MD 9.28, 95% CI 7.52-11.04, P<.001; I2=0%), FIM (MD 13.60, 95% CI 9.55-17.64, P<.001; I2=72%; or MD 9.52, 95% CI 7.02-12.02, P<.001; I2=4% after the removal of outliers [21,25]), WMFT (MD 13.94, 95% CI 8.24-19.64, P<.001; I2=82%), and BBT (MD 8.60, 95% CI 4.54-12.67, P<.001; I2=21%; Figure 3).

    Subgroup Analyses

    Age Influencing the VR Effect

    Compared with those in the CON group, younger patients (n=301, mean age 56.45 years) in the VRT group demonstrated significantly greater improvements in upper limb function measured by the FMA-UE (MD 5.34, 95% CI 2.18-8.5, P<.001; Table 4). No statistically significant heterogeneity was observed among older patients (SMD 1.96, 95% CI −1.50 to 5.41, P=.27; Table 4). Similarly, younger patients showed greater improvements in independence in activities of daily living measured by the FIM (MD 6.55, 95% CI 2.36-10.75, P=.002), while no statistically significant heterogeneity was observed among older patients (MD 2.68, 95% CI −0.47 to 5.83, P=.10; Table 5).

    Table 4. Subgroup analyses on age, total hours, trial length, and duration based on FMA-UEa measures.
    Outcomes, moderating factors, and between-group/within-group comparisonNumber of trials analyzed and number of participants involvedMean difference (95% CI)P valueI2 (%)
    FMA-UE




    		Age (years)






    		Younger (≤60.52 years)







    		Between-group3; N (VRTb)=301, N (CONc)=1505.34 (2.18 to 8.50).00090



    		Within-group4; N=3526.96 (4.64 to 9.29)<.00148


    		Older (>60.52 years)







    		Between-group5; N (VRT)=154, N (CON)=1511.96 (–1.50 to 5.41).270



    		Within-group5; N=15411.93 (8.54 to 15.33)<.00142

    		Total dosage (hours)






    		<15







    		Between-group2; N (VRT)=45, N (CON)=450.79 (–4.90 to 6.47).790



    		Within-group2; N=4513.90 (8.09 to 19.71)<.0010


    		≥15 and ≤30







    		Between-group3; N (VRT)=56, N (CON)=549.67 (4.19 to 15.15).00050



    		Within-group4; N=10711.14 (2.99 to 19.30)<.00172


    		>30







    		Between-group3; N (VRT)=354, N (CON)=2022.95 (0.06 to 5.84).050



    		Within-group3; N=3547.07 (4.76 to 9.38)<.0010

    		Trial length (weeks)






    		<4







    		Between-group2; N (VRT)=45, N (CON)=450.79 (–4.90 to 6.47).790



    		Within-group2; N=4513.90 (8.09 to 19.71)<.0010


    		≥4 and ≤6







    		Between-group5; N (VRT)=392, N (CON)=2394.02 (1.39 to 6.65).00315



    		Within-group5; N=39210.88 (6.07 to 15.68)<.00167


    		>6







    		Between-group1; N (VRT)=18, N (CON)=1711.0 (0.13 to 21.87).05N/Ad



    		Within-group2; N=694.15 (–1.78 to 10.09).170

    		Duration of a session (hours)






    		<1







    		Between-group2; N (VRT)=45, N (CON)=450.79 (–4.90 to 6.47).790



    		Within-group2; N=4513.90 (8.09 to 19.71)<.0010


    		≥1 and<2







    		Between-group3; N (VRT)=56, N (CON)=549.67 (4.19 to 15.15).00050



    		Within-group4; N=10711.14 (2.99 to 19.30)<.00172


    		≥2








    		Between-group3; N (VRT)=354, N (CON)=2022.95 (0.06 to 5.84).050



    		Within-group3; N=3547.07 (4.76 to 9.38)<.0010

    		Frequency (days/week)






    		≤4







    		Between-group3; N (VRT)=61, N (CON)=597.01 (–1.67 to 15.68).1158



    		Within-group4; N=11211.17 (3.14 to 19.20).00673


    		>4







    		Between-group5; N (VRT)=394, N (CON)=2423.48 (0.87 to 6.09).0090



    		Within-group5; N=3947.86 (5.69 to 10.03)<.00127

    aFMA-UE: Upper Extremity Fugl-Meyer Assessment.

    bVRT: virtual reality therapy.

    cCON: conventional therapy.

    dN/A: not applicable.

    Table 5. Subgroup analyses on age, total hours, trial length, and duration based on the Functional Independence Measure.
    Outcomes, moderating factors, and between-group/within-group comparisonNumber of trials analyzed and number of participants involvedMean difference (95% CI)P valueI2 (%)
    Functional Independence Measure




    		Age (years)





    		Younger (≤60.52)






    		Between-group2; N (VRTa)=283, N (CONb)=1336.55 (2.36 to 10.75).00263



    		Within-group2; N=28311.54 (3.91 to 19.17).00392


    		Older (>60.52)






    		Between-group5; N (VRT)=205, N (CON)=1992.68 (–0.47 to 5.83).1026



    		Within-group5; N=20515.13 (11.72 to 18.55)<.00136

    		Total dosage (hours)





    		<30






    		Between-group2; N (VRT)=87, N (CON)=832.07 (–4.65 to 8.79).5560



    		Within-group2; N=8717.52 (9.29 to 25.75)<.00171


    		≥30 and <40






    		Between-group2; N (VRT)=47, N (CON)=475.06 (2.06 to 8.07).00111



    		Within-group2; N=4715.28 (12.62 to 17.94)<.0010


    		≥40






    		Between-group3; N (VRT)=354, N (CON)=2026.70 (3.46 to 9.94)<.00146



    		Within-group3; N=3548.43 (5.6 to 11.26)<.0010

    		Duration of a session (hours)





    		<1






    		Between-group1; N (VRT)=25, N (CON)=255.90 (–0.96 to 12.76).09N/Ac



    		Within-group1; N=2521.80 (15.33 to 28.27)<.001N/A


    		≥1 and <2






    		Between-group3; N (VRT)=109, N (CON)=1053.54 (–1.60 to 8.68).1861



    		Within-group3; N=10914.98 (12.54 to 17.41)<.0010


    		≥2






    		Between-group3; N (VRT)=354, N (CON)=2026.70 (3.46 to 9.94)<.00146



    		Within-group3; N=3548.43 (5.60 to 11.26)<.0010

    		Frequency (days/week)





    		≤4






    		Between-group2; N (VRT)=87, N (CON)=832.07 (–4.65 to 8.79).5560



    		Within-group2; N=8717.52 (9.29 to 25.75)<.00171


    		>4






    		Between-group5; N (VRT)=401, N (CON)=2495.82 (3.62 to 8.03)<.00126



    		Within-group5; N=40111.89 (7.31 to 16.47)<.001 71

    aVRT: virtual reality therapy.

    bCON: conventional therapy.

    cN/A: not applicable.

    Dosage (Total Hours) Influencing the VR Effect

    Comparisons were conducted between studies with total intervention delivery times of 15 hours or less and those exceeding 15 hours. Among studies with less than 15 hours of intervention, the VRT group did not show significant changes in upper limb function, as measured by the FMA-UE, compared with the CON group before and after the intervention (MD 0.79, 95% CI −4.90 to 6.47, P=.79; Table 4). By contrast, among studies with over 15 hours of intervention, the VRT group demonstrated significant improvements in upper limb function measured by the FMA-UE (MD 9.67, 95% CI 4.19-15.15, P<.001; Table 4).

    Trial Length Influencing the VR Effect

    When the trial duration was 4 weeks or longer, improvements in upper limb function measured by the FMA-UE in the VRT group were greater than those in the CON group (MD 4.02, 95% CI 1.39-6.65, P<.001; Table 4). For within-group comparisons, trials lasting over 4 weeks showed significant improvements in upper limb function in the VRT group measured by the ARAT (MD 9.49, 95% CI 7.69-11.29, P<.001; Table 6) and the BBT (MD 7.43, 95% CI 0.50-14.36, P=.04; Table 7). The VRT group also demonstrated improvements in upper limb function measured by the FMA-UE when the trial duration was less than 6 weeks (MD 10.88, 95% CI 6.07-15.68, P<.001; Table 4), but no significant improvements were observed when the trial duration was 6 weeks or longer (MD 4.15, 95% CI −1.78 to 10.09, P=.17; Table 4).

    Table 6. Subgroup analyses on age, total hours, trial length, and duration based on Action Research Arm Test measures.
    Outcomes and moderating factorsBetween-group/within-group comparisonNumber of trials analyzed and number of participants involvedMean difference (95% CI)P valueI2 (%)
    Action Research Arm Test





    		Age (years)






    		Younger (≤60.52)Within-group2; N=369.13 (7.28 to 10.98)<.00133


    		Older (>60.52)Within-group2; N=8210.73 (4.92 to 16.54)<.0010

    		Total dosage (hours)






    		<30Within-group2; N=8210.73 (4.92 to 16.54)<.0010


    		≥30Within-group2; N=369.13 (7.28 to 10.98)<.00133

    		Trial length(weeks)






    		<4Within-group2; N=364.32 (–4.39 to 13.04).330


    		≥4Within-group2; N=829.49 (7.69 to 11.29)<.0010

    		Duration of a session (hours)






    		≤1Within-group2; N=8210.73 (4.92 to 16.54)<.0010


    		>1Within-group2; N=369.13 (7.28 to 10.98)<.00133

    		Frequency (days/week)






    		≤4Within-group2; N=829.49 (7.69 to 11.29)<.0010


    		>4Within-group2; N=364.32 (–4.39 to 13.04).33 0
    Table 7. Subgroup analyses on age, total hours, trial length, and duration based on Box-Block Test measures.
    Outcomes, moderating factors, and between-group/within-group comparisonNumber of trials analyzed and number of participants involvedMean difference (95% CI)P valueI2 (%)
    Box-Block Test




    		Age (years)





    		Younger (≤60.52)






    		Between-group1; N (VRTa)=18, N (CONb)=177.70 (–5.71 to 21.11).26N/Ac



    		Within-group1; N=182.90 (–9.36 to 15.16).64N/A


    		Older (>60.52)






    		Between-group3; N (VRT)=102, N (CON)=1072.72 (–1.96 to 7.40).250



    		Within-group3; N=1029.31 (5.00 to 13.62)<.00130

    		Total dosage (hours)





    		≤15






    		Between-group2; N (VRT)=40, N (CON)=495.42 (–2.49 to 13.32).180



    		Within-group2; N=402.33 (–5.32 to 9.98).550


    		>15






    		Between-group2; N (VRT)=80, N (CON)=752.28 (–3.05 to 7.62).40



    		Within-group2; N=8011.07 (6.27 to 15.87)<.0010

    		Trial length (weeks)





    		≤4






    		Between-group2; N (VRT)=84, N (CON)=902.04 (–3.53 to 7.60).470



    		Within-group2; N=847.70 (–1.80 to 17.20).1165


    		>4






    		Between-group2; N (VRT)=36, N (CON)=345.36 (–1.92 to 12.63).150



    		Within-group2; N=367.43 (0.50 to 14.36).040

    		Duration of a session (hours)





    		<1






    		Between-group2; N (VRT)=40, N (CON)=495.42 (–2.49 to 13.32).180



    		Within-group2; N=402.33 (–5.32 to 9.98).550


    		≥1






    		Between-group2; N (VRT)=80, N (CON)=752.28 (–3.05 to 7.62).40



    		Within-group2; N=8011.07 (6.27 to 15.87)<.0010

    		Frequency (days/week)





    		<4






    		Between-group2; N (VRT)=36, N (CON)=345.36 (–1.92 to 12.63).150



    		Within-group2; N=367.43 (0.50 to 14.36).040


    		≥4






    		Between-group2; N (VRT)=84, N (CON)=902.04 (–3.53 to 7.60).470



    		Within-group2; N=847.70 (–1.80 to 17.20).11 65

    aVRT: virtual reality therapy.

    bCON: conventional therapy.

    cN/A: not applicable.

    Duration (Hour/Session) and Frequency (Session/Week) Influencing the VR Effect

    For session durations shorter than 2 hours, participants in the VRT group (mean 1.17 hours) demonstrated significantly greater improvements in upper limb function, as measured by the FMA-UE, compared with those in the CON group (mean 1.17 hours) (MD 9.67, 95% CI 4.19-15.15, P<.001; Table 4). For session durations equal to or longer than 2 hours (ie, ≥2 hours/session), participants in the VRT group (mean 2.0 hours) also showed significantly greater improvements in independence in activities of daily living, as measured by the FIM, compared with those in the CON group (mean 2.0 hours; MD 6.70, 95% CI 3.46-9.94, P<.001; Table 5).

    When the frequency exceeded 4 sessions per week, improvements in upper limb function measured by the FMA-UE (MD 3.48, 95% CI 0.87-6.09, P<.001) and independence in activities of daily living measured by the FIM in the VRT group were greater than those in the CON group (MD 5.82, 95% CI 3.62-8.03, P<.001) (Tables 4 and 5). Conversely, when the frequency was 4 or fewer sessions per week, significant changes in upper limb function measured by the ARAT were observed in the VRT group (MD 9.49, 95% CI 7.69-11.29, P<.001), but no significant changes were found when the frequency exceeded 4 sessions per week (MD 4.32, 95% CI –4.39 to 13.04, P=.33; Table 6).

    Content Design Influencing the VR Effect

    The content design factor was defined based on the National Clinical Guideline for Stroke for the United Kingdom and Ireland (2023 edition) [19]. The number of features in the VR interventions across all studies ranged from 2 to 10. Comparisons were made between studies with a fewer number of content features (n=7, features ≤5) and those with a greater number of features (n=8, features>5; Table 8). The common outcome variable among the included studies was upper limb function measured by the FMA-UE, and thus the comparisons were based on this measure. The MDs in FMA-UE scores did not significantly differ between the fewer and greater feature subgroups (MD 3.89, 95% CI –6.40 to 1.09; effect Z=1.36, P=.16; Figure 4; see also [11,42-46,48,50]).

    Table 8. Content analysis of the design features of the VRT used in the studies.
    StudiesRelative quantity of content featuresaContent feature items
    1b2c3d4e5f6g7h8i9j10k
    Brunner et al [40]+N/Al✓m✓N/AN/AN/AN/AN/A✓N/A
    Abd El-Kafy et al [41]+++✓✓✓✓✓N/A✓N/AN/AN/A
    Gueye et al [42]+N/A✓✓N/AN/AN/AN/A✓N/AN/A
    Hsu et al [11]✓N/A✓✓N/AN/AN/AN/AN/AN/AN/A
    Kiper et al [43]+++✓✓✓✓✓✓✓✓N/A✓
    Kiper et al [44]+++✓✓✓✓✓✓✓✓N/A✓
    Rong et al [45]++✓✓✓N/AN/AN/A✓N/AN/AN/A
    Schuster-Amft et al [16]+++✓✓✓✓✓N/A✓✓✓N/A
    Sin and Lee [46]++N/A✓✓✓✓N/A✓N/AN/AN/A
    Taveggia et al [47]+N/AN/A✓✓N/AN/AN/AN/AN/AN/A
    Yao et al [48]+++N/AN/A✓✓✓✓✓N/AN/A✓
    Zheng et al [49]+++✓✓✓✓✓N/A✓✓✓N/A
    Turolla et al [50]+++✓✓✓N/A✓✓✓N/A✓N/A
    Borstad et al [33]+++✓✓✓✓✓✓✓✓N/A✓
    Sebastián-Romagosa et al [34]+N/A✓✓N/AN/AN/AN/AN/AN/AN/A

    aThe ‘+’, ‘++’, and ‘+++’ keys refer to a few (2-3), medium (4-5), and many (≥6) numbers of features, respectively.

    bItem 1: Offering exercise and functional tasks.

    cItem 2: Having individualized goals.

    dItem 3: Quantifying activity.

    eItem 4: Considering comorbidities and baseline activity level.

    fItem 5: Incorporating patient’s needs.

    gItem 6: Matching patient’s background such as education level.

    hItem 7: Patient’s goal and interest directed.

    iItem 8: Involving goal setting.

    jItem 9: Progressive difficulty level.

    kItem 10: Promoting self-efficacy.

    lN/A: not applicable.

    mThe symbol ‘✓’ indicates that descriptions of the virtual reality intervention are found to contain the corresponding feature item.

    ‎
    Figure 4. Forest plot of comparisons of VRT content design between the fewer and a greater number of feature subgroups. VRT: virtual reality therapy.

    Principal Findings

    This study aimed to examine the influence of patients’ age, delivery dosage and schedule, and content design on the outcomes of virtual reality interventions for patients with stroke. The meta-analytic results revealed that virtual reality training had significantly greater therapeutic effects than conventional training in improving upper limb function in patients with stroke. These effects included enhancements in upper limb function, as measured by the FMA-UE and ARAT, and improvements in activities of daily living, as measured by the FIM.

    The demographic factor associated with more significant therapeutic effects was younger patient age compared with older age (mean 56.5 vs 65.9 years). Virtual reality training content factors contributing to greater effects included higher training dosage (>15 hours) delivered over 4-6 weeks and shorter sessions (mean 1.2 vs 2.0 hours) scheduled 4 or more times per week. Notably, shorter sessions (mean 1.2 hours) were linked to better outcomes in upper limb function, whereas longer sessions (mean 2.0 hours) were associated with improved outcomes in activities of daily living.

    Our findings on the therapeutic effects of virtual reality training are largely consistent with those reported in 4 previous meta-analytic studies [9,36,54,55]. The consistent results confirm that virtual reality, as a novel and potentially useful technology, is effective in improving upper limb function and activities of daily living compared with conventional therapy after stroke. However, an inconsistent result was observed in the nonsignificant effects of virtual reality training on improving patients’ upper limb dexterity (measured by the BBT). This nonsignificant finding could be attributed to the smaller number of studies that used finger dexterity as an outcome variable (n=4) compared with those focusing on upper limb function or activities of daily living (n=8).

    Younger patients with stroke demonstrated better outcomes than older patients in both upper limb function and activities of daily living. One plausible explanation is that younger patients may have been more motivated and actively engaged in the training, leading to better results. Previous studies have shown that increasing age is associated with lower participation in cognitively demanding activities [56]. Reduced participation inevitably results in lower gains from training. Spiteri et al [57] found that older adults (eg, 65-70 years) faced barriers such as mobility challenges, limited community accessibility, and insufficient guidance from health care professionals, which hindered their participation in physical activity (also see [58]). Our findings further emphasize the value of virtual reality training in poststroke rehabilitation. Addressing potential barriers, such as ensuring adequate supervision from health care professionals and improving accessibility, could enhance the effectiveness of VR-based training for older patients.

    VR-based training offers immersive, imaginative, and interactive experiences [59,60] that facilitate functional recovery in patients with stroke [61,62]. These vivid experiences have been linked to participants engaging in more intensive cognitive and motor processes during training [63-66]. The cognitive processes involve bottom-up attention, which entails the encoding and processing of visual and auditory stimuli emitted in the virtual environment [67,68]. Top-down attention, by contrast, involves cognitive control, decision-making, and motor planning to produce motor responses [69,70]. The content design of virtual reality training did not significantly influence its therapeutic effects. In this study, we adhered to the principles outlined by the United Kingdom and Ireland for designing therapy aimed at promoting functional recovery in patients with stroke. The number of different features, whether fewer or more, based on these principles, did not emerge as a significant factor. However, 2 content features shared by most studies were setting individualized goals (item 2) and quantifying activities (item 3). Our findings suggest that counting repetitions of movements or responses performed by patients and setting treatment goals tailored to individual needs may be essential or sufficient features for designing effective virtual reality interventions.

    The cognitive and motor processes involved in virtual reality training need to be repetitive to drive functional changes in patients with stroke. Our results suggest that more than 15 hours of training, scheduled over 4-6 weeks with 4 or more 1-hour sessions per week, can maximize improvements in upper limb function. By contrast, achieving significant gains in independence in activities of daily living appears to require a higher dosage, with 30 or more hours of training needed to produce meaningful outcomes. Repetitiveness is essential for initiating neural changes after a stroke. At the neurological level, posttraining changes in the motor cortices involve alterations in functional connectivity within neural networks, such as the prefrontal cortex and the basal ganglia [71,72]. Recovery of upper extremity function after a stroke has been linked to changes in functional connectivity between the bilateral primary motor cortex (M1) and the dorsolateral prefrontal cortex [73,74]. Additionally, repetitive training leads to the functional reorganization of the sensorimotor networks and the extrapyramidal system [75].

    Limitation

    This study has several limitations. First, the subgroup analyses did not include outcomes measured by the ARAT or the WMFT, limiting the generalizability of the results. Caution is therefore needed when interpreting the findings. Second, the therapeutic effects of VR-based training were derived from outcomes measured shortly after the completion of training. As such, the findings do not extend to short- or long-term posttraining effects. Evidence regarding the long-term benefits of VR-enhanced exercise training for improving upper limb function in patients with stroke remains limited. However, a study examining the long-term treatment effects of VR in cardiovascular rehabilitation found that combining VR-based therapy with conventional treatment was associated with sustained long-term effects on hemodynamic and autonomic outcomes. This suggests that VR could serve as a valuable new treatment modality when integrated with cardiovascular rehabilitation sessions [51-53,76]. Standalone VR treatment lacks sensory and proprioceptive inputs to the central nervous system, limiting the functionality of the central-peripheral loop. Future longitudinal studies should explore the combination of VR-based therapy with actual exercises, focusing on how to integrate these approaches effectively for patients with stroke.

    Third, significant heterogeneity (I2>50%) was observed in the meta-analysis. In the whole-group analysis, participants in the VRT group showed significantly greater improvement than those in the CON group, with I2<50%, indicating that the results from the studies were consistent in favoring VRT. However, in the within-group pre- and postintervention analysis, we found that some studies had I2>50%, such as the WMFT (I2=82%). This could be attributed to clinical variability in intervention delivery, including factors such as dosage, trial length, and frequency. The heterogeneity of participants across the included trials likely reduced the effect sizes of the observed therapeutic effects, particularly for older patients and improvements in finger dexterity. Similarly, the relatively small number of studies in the subgroup analyses may have reduced the power to identify training design content as a significant outcome factor. Future research should include more studies and patients with stroke in the analyses and investigate the long-term effects of virtual reality training.

    Conclusions

    Virtual reality training was effective in promoting upper limb function and independence in activities of daily living for patients with stroke. Patient age and the dosage and schedule of training delivery are factors that influence its therapeutic effects. Younger patients who participate in training for more than 15 hours tend to show better outcomes compared with older patients. Training content delivered in 1-hour sessions, with more than 4 sessions per week over 4-6 weeks, could further maximize the therapeutic effects. Additionally, the design of the VR content should be tailored to the individual needs of the patients and involve them in setting treatment goals.

    Acknowledgments

    This work was supported by the Wuxi Taihu Lake Talent Development Program in Medical and Health, funded by the Wuxi Municipal Office of Talent Affairs (grant THRC23074004); the Healthcare Commission Foundation of Wuxi City (grant M202123); the Wuxi Ninth People’s Hospital Affiliated with Soochow University projects (grants QD202102 and YB202103); and the Wuxi Philosophy and Social Sciences Project (grant WXSK21-JY-B04).

    No generative artificial intelligence was used in any portion of the manuscript writing.

    Data Availability

    All data generated or analyzed during this study are included in this published article. Further inquiries can be directed to the corresponding author.

    Authors' Contributions

    SX and YX contributed equally and are joint first authors. Both YX and CCHC are corresponding authors. YX's email address is xudeng2@163.com.

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    Search strategy.

    PDF File (Adobe PDF File), 149 KB
    Multimedia Appendix 2

    PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist.

    PDF File (Adobe PDF File), 95 KB
    1. Johnson W, Onuma O, Owolabi M, Sachdev S. Stroke: a global response is needed. Bull World Health Organ. Sep 01, 2016;94(9):634-634A. [CrossRef]
    2. GBD 2016 Lifetime Risk of Stroke Collaborators T. Global, regional, and country-specific lifetime risks of stroke, 1990 and 2016. N Engl J Med. Dec 20, 2018;379(25):2429-2437. [CrossRef]
    3. Bernhardt J, Urimubenshi G, Gandhi DBC, Eng JJ. Stroke rehabilitation in low-income and middle-income countries: a call to action. The Lancet. Oct 2020;396(10260):1452-1462. [CrossRef]
    4. Langhammer B, Stanghelle JK. Bobath or motor relearning programme? A comparison of two different approaches of physiotherapy in stroke rehabilitation: a randomized controlled study. Clin Rehabil. Aug 01, 2000;14(4):361-369. [CrossRef] [Medline]
    5. Logigian M, Samuels M, Falconer J, Zagar R. Clinical exercise trial for stroke patients. Arch Phys Med Rehabil. Aug 1983;64(8):364-367. [Medline]
    6. van Vliet PM, Lincoln NB, Foxall A. Comparison of Bobath based and movement science based treatment for stroke: a randomised controlled trial. J Neurol Neurosurg Psychiatry. Apr 2005;76(4):503-508. [FREE Full text] [CrossRef] [Medline]
    7. Dimyan MA, Cohen LG. Neuroplasticity in the context of motor rehabilitation after stroke. Nat Rev Neurol. Feb 18, 2011;7(2):76-85. [FREE Full text] [CrossRef] [Medline]
    8. Tieri G, Morone G, Paolucci S, Iosa M. Virtual reality in cognitive and motor rehabilitation: facts, fiction and fallacies. Expert Rev Med Devices. Feb 10, 2018;15(2):107-117. [CrossRef] [Medline]
    9. Saposnik G, Levin M. Virtual reality in stroke rehabilitation. Stroke. May 2011;42(5):1380-1386. [CrossRef]
    10. Alves SS, Ocamoto GN, de Camargo PS, Santos ATS, Terra AMSV. Effects of virtual reality and motor imagery techniques using Fugl Meyer Assessment scale in post-stroke patients. International Journal of Therapy and Rehabilitation. Nov 02, 2018;25(11):587-596. [CrossRef]
    11. Hsu H, Kuo L, Lin Y, Su F, Yang T, Lin C. Effects of a virtual reality-based mirror therapy program on improving sensorimotor function of hands in chronic stroke patients: a randomized controlled trial. Neurorehabil Neural Repair. Jun 28, 2022;36(6):335-345. [CrossRef] [Medline]
    12. Ballester BR, Nirme J, Camacho I, Duarte E, Rodríguez S, Cuxart A, et al. Domiciliary VR-based therapy for functional recovery and cortical reorganization: randomized controlled trial in participants at the chronic stage post stroke. JMIR Serious Games. Aug 07, 2017;5(3):e15. [FREE Full text] [CrossRef] [Medline]
    13. Buccellato K, Nordstrom M, Murphy J, Burdea G, Polistico K, House G, et al. A randomized feasibility trial of a novel, integrative, and intensive virtual rehabilitation program for service members post-acquired brain injury. Mil Med. Feb 12, 2020;185(1-2):e203-e211. [CrossRef] [Medline]
    14. Ikbali Afsar S, Mirzayev I, Umit Yemisci O, Cosar Saracgil SN. Virtual reality in upper extremity rehabilitation of stroke patients: a randomized controlled trial. J Stroke Cerebrovasc Dis. Dec 2018;27(12):3473-3478. [CrossRef] [Medline]
    15. McEwen D, Taillon-Hobson A, Bilodeau M, Sveistrup H, Finestone H. Virtual reality exercise improves mobility after stroke. Stroke. Jun 2014;45(6):1853-1855. [CrossRef]
    16. Schuster-Amft C, Eng K, Suica Z, Thaler I, Signer S, Lehmann I, et al. Effect of a four-week virtual reality-based training versus conventional therapy on upper limb motor function after stroke: a multicenter parallel group randomized trial. PLoS One. Oct 24, 2018;13(10):e0204455. [FREE Full text] [CrossRef] [Medline]
    17. Sánchez-Herrera-Baeza P, Cano-de-la-Cuerda R, Oña-Simbaña ED, Palacios-Ceña D, Pérez-Corrales J, Cuenca-Zaldivar JN, et al. The impact of a novel immersive virtual reality technology associated with serious games in Parkinson’s disease patients on upper limb rehabilitation: a mixed methods intervention study. Sensors. Apr 11, 2020;20(8):2168. [FREE Full text] [CrossRef] [Medline]
    18. Pazzaglia C, Imbimbo I, Tranchita E, Minganti C, Ricciardi D, Lo Monaco R, et al. Comparison of virtual reality rehabilitation and conventional rehabilitation in Parkinson's disease: a randomised controlled trial. Physiotherapy. Mar 2020;106:36-42. [CrossRef] [Medline]
    19. Intercollegiate Stroke Working Party. National Clinical Guideline for Stroke for the UK and Ireland. Intercollegiate Stroke Working Party. May 4, 2023. URL: https://www.strokeguideline.org [accessed 2025-01-28]
    20. Page SJ, Schmid A, Harris JE. Optimizing terminology for stroke motor rehabilitation: recommendations from the American Congress of Rehabilitation Medicine Stroke Movement Interventions Subcommittee. Arch Phys Med Rehabil. Aug 21, 2012;93(8):1395-1399. [FREE Full text] [CrossRef] [Medline]
    21. Nudo RJ, Wise BM, SiFuentes F, Milliken GW. Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct. Science. Jun 21, 1996;272(5269):1791-1794. [CrossRef] [Medline]
    22. Krakauer JW, Carmichael ST, Corbett D, Wittenberg GF. Getting neurorehabilitation right: what can be learned from animal models? Neurorehabil Neural Repair. Oct 30, 2012;26(8):923-931. [FREE Full text] [CrossRef] [Medline]
    23. Kleim JA, Barbay S, Cooper NR, Hogg TM, Reidel CN, Remple MS, et al. Motor learning-dependent synaptogenesis is localized to functionally reorganized motor cortex. Neurobiol Learn Mem. Jan 2002;77(1):63-77. [CrossRef] [Medline]
    24. Luke LM, Allred RP, Jones TA. Unilateral ischemic sensorimotor cortical damage induces contralesional synaptogenesis and enhances skilled reaching with the ipsilateral forelimb in adult male rats. Synapse. Dec 15, 2004;54(4):187-199. [CrossRef] [Medline]
    25. Birkenmeier RL, Prager EM, Lang CE. Translating animal doses of task-specific training to people with chronic stroke in 1-hour therapy sessions: a proof-of-concept study. Neurorehabil Neural Repair. Sep 27, 2010;24(7):620-635. [FREE Full text] [CrossRef] [Medline]
    26. Carey JR, Durfee WK, Bhatt E, Nagpal A, Weinstein SA, Anderson KM, et al. Comparison of finger tracking versus simple movement training via telerehabilitation to alter hand function and cortical reorganization after stroke. Neurorehabil Neural Repair. 2007;21(3):216-232. [CrossRef] [Medline]
    27. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. Mar 29, 2021;372:n71-112. [FREE Full text] [CrossRef] [Medline]
    28. Maher C, Sherrington C, Herbert R, Moseley A, Elkins M. Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther. 2003;83(8):713-721. [CrossRef]
    29. Slim K, Nini E, Forestier D, Kwiatkowski F, Panis Y, Chipponi J. Methodological index for non-randomized studies (minors): development and validation of a new instrument. ANZ J Surg. Sep 08, 2003;73(9):712-716. [CrossRef] [Medline]
    30. Cashin AG, McAuley JH. Clinimetrics: Physiotherapy Evidence Database (PEDro) Scale. J Physiother. Jan 2020;66(1):59. [FREE Full text] [CrossRef] [Medline]
    31. Luo D, Wan X, Liu J, Tong T. Optimally estimating the sample mean from the sample size, median, mid-range, and/or mid-quartile range. Stat Methods Med Res. Sep 27, 2016;27(6):1785-1805. [CrossRef] [Medline]
    32. Higgins J, Li T, Deeks J. Cochrane Handbook for Systematic Reviews of Interventions. London, UK. Cochrane; Sep 20, 2019.
    33. Borstad AL, Crawfis R, Phillips K, Lowes LP, Maung D, McPherson R, et al. In-home delivery of constraint-induced movement therapy via virtual reality gaming. J Patient Cent Res Rev. Jan 30, 2018;5(1):6-17. [CrossRef] [Medline]
    34. Sebastián-Romagosa M, Cho W, Ortner R, Murovec N, Von Oertzen T, Kamada K, et al. Brain computer interface treatment for motor rehabilitation of upper extremity of stroke patients-a feasibility study. Front Neurosci. Oct 21, 2020;14:591435. [FREE Full text] [CrossRef] [Medline]
    35. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol. Dec 19, 2014;14(1):135. [FREE Full text] [CrossRef] [Medline]
    36. Laver KE, Lange B, George S, Deutsch J, Saposnik G, Crotty M. Virtual reality for stroke rehabilitation. Cochrane Database Syst Rev. Nov 20, 2017;11(11):CD008349. [FREE Full text] [CrossRef] [Medline]
    37. Mekbib DB, Zhao Z, Wang J, Xu B, Zhang L, Cheng R, et al. Proactive motor functional recovery following immersive virtual reality-based limb mirroring therapy in patients with subacute stroke. Neurotherapeutics. Oct 2020;17(4):1919-1930. [CrossRef] [Medline]
    38. Zhang J, Fong K, Welage N, Liu K. The activation of the mirror neuron system during action observation and action execution with mirror visual feedback in stroke: a systematic review. Neural Plast. Apr 24, 2018;2018:2321045. [FREE Full text] [CrossRef] [Medline]
    39. Chen J, Or CK, Chen T. Effectiveness of using virtual reality–supported exercise therapy for upper extremity motor rehabilitation in patients with stroke: systematic review and meta-analysis of randomized controlled trials. J Med Internet Res. Jun 20, 2022;24(6):e24111. [FREE Full text] [CrossRef] [Medline]
    40. Brunner I, Skouen JS, Hofstad H, Aßmus J, Becker F, Sanders A, et al. Virtual reality training for upper extremity in subacute stroke (VIRTUES). Neurology. Dec 12, 2017;89(24):2413-2421. [CrossRef]
    41. Abd El-Kafy EM, Alshehri MA, El-Fiky AA, Guermazi MA, Mahmoud HM. The effect of robot-mediated virtual reality gaming on upper limb spasticity poststroke: a randomized-controlled trial. Games Health J. Apr 01, 2022;11(2):93-103. [CrossRef] [Medline]
    42. Gueye T, Dedkova M, Rogalewicz V, Grunerova-Lippertova M, Angerova Y. Early post-stroke rehabilitation for upper limb motor function using virtual reality and exoskeleton: equally efficient in older patients. Neurol Neurochir Pol. Feb 28, 2021;55(1):91-96. [CrossRef]
    43. Kiper P, Agostini M, Luque-Moreno C, Tonin P, Turolla A. Reinforced feedback in virtual environment for rehabilitation of upper extremity dysfunction after stroke: preliminary data from a randomized controlled trial. Biomed Res Int. Mar 13, 2014;2014:752128. [FREE Full text] [CrossRef] [Medline]
    44. Kiper P, Szczudlik A, Agostini M, Opara J, Nowobilski R, Ventura L, et al. Virtual reality for upper limb rehabilitation in subacute and chronic stroke: a randomized controlled trial. Arch Phys Med Rehabil. May 2018;99(5):834-842.e4. [CrossRef] [Medline]
    45. Rong J, Ding L, Xiong L, Zhang W, Wang W, Deng M, et al. Mirror visual feedback prior to robot-assisted training facilitates rehabilitation after stroke: a randomized controlled study. Front Neurol. Jul 8, 2021;12:683703. [FREE Full text] [CrossRef] [Medline]
    46. 46Sin H, Lee G. Additional virtual reality training using Xbox Kinect in stroke survivors with hemiplegia. American Journal of Physical Medicine and Rehabilitation. 2013;92(10):871-880. [CrossRef]
    47. Taveggia G, Borboni A, Salvi L, Mulé C, Fogliaresi S, Villafañe J. Efficacy of robot-assisted rehabilitation for the functional recovery of the upper limb in post-stroke patients: a randomized controlled study. European Journal of Physical and Rehabilitation Medicine. 2016;52(6):767-773. [CrossRef]
    48. Yao X, Cui L, Wang J, Feng W, Bao Y, Xie Q. Effects of transcranial direct current stimulation with virtual reality on upper limb function in patients with ischemic stroke: a randomized controlled trial. J Neuroeng Rehabil. Jun 15, 2020;17(1):73. [CrossRef] [Medline]
    49. Zheng C, Liao W, Xia W. Effect of combined low-frequency repetitive transcranial magnetic stimulation and virtual reality training on upper limb function in subacute stroke: a double-blind randomized controlled trail. J Huazhong Univ Sci Technolog Med Sci. Apr 16, 2015;35(2):248-254. [CrossRef] [Medline]
    50. Turolla A, Dam M, Ventura L, Tonin P, Agostini M, Zucconi C, et al. Virtual reality for the rehabilitation of the upper limb motor function after stroke: a prospective controlled trial. J Neuroeng Rehabil. Aug 01, 2013;10(1):85. [FREE Full text] [CrossRef] [Medline]
    51. Ersoy C, Iyigun G. Boxing training in patients with stroke causes improvement of upper extremity, balance, and cognitive functions but should it be applied as virtual or real? Topics in Stroke Rehabilitation. Jun 23, 2020;28(2):112-126. [CrossRef]
    52. Junior VADS, Santos MDS, Ribeiro NMDS, Maldonado IL. Combining proprioceptive neuromuscular facilitation and virtual reality for improving sensorimotor function in stroke survivors: a randomized clinical trial. J Cent Nerv Syst Dis. Jul 25, 2019;11:1179573519863826. [CrossRef] [Medline]
    53. Kizony R, Weiss PL, Harel S, Feldman Y, Obuhov A, Zeilig G, et al. Tele-rehabilitation service delivery journey from prototype to robust in-home use. Disabil Rehabil. Jul 22, 2017;39(15):1532-1540. [CrossRef] [Medline]
    54. Lohse KR, Hilderman CGE, Cheung KL, Tatla S, Van der Loos HFM. Virtual reality therapy for adults post-stroke: a systematic review and meta-analysis exploring virtual environments and commercial games in therapy. PLoS One. Mar 28, 2014;9(3):e93318. [FREE Full text] [CrossRef] [Medline]
    55. Aminov A, Rogers JM, Middleton S, Caeyenberghs K, Wilson PH. What do randomized controlled trials say about virtual rehabilitation in stroke? A systematic literature review and meta-analysis of upper-limb and cognitive outcomes. J NeuroEngineering Rehabil. Mar 27, 2018;15(1):29. [CrossRef]
    56. Hess TM. Selective engagement of cognitive resources: motivational influences on older adults' cognitive functioning. Perspect Psychol Sci. Jul 16, 2014;9(4):388-407. [FREE Full text] [CrossRef] [Medline]
    57. Spiteri K, Broom D, Bekhet A, de Caro JX, Laventure B, Grafton K. Barriers and motivators of physical activity participation in middle-aged and older-adults - a systematic review. J Aging Phys Act. Sep 01, 2019;27(4):929-944. [CrossRef] [Medline]
    58. Gingrich N, Bosancich J, Schmidt J, Sakakibara BM. Capability, opportunity, motivation, and social participation after stroke. Top Stroke Rehabil. Jul 05, 2023;30(5):423-435. [CrossRef] [Medline]
    59. Subramanian SK, Prasanna SS. Virtual reality and noninvasive brain stimulation in stroke: how effective is their combination for upper limb motor improvement?-a meta-analysis. PM R. Nov 2018;10(11):1261-1270. [CrossRef] [Medline]
    60. Kim W, Cho S, Park S, Lee J, Kwon S, Paik N. A low cost kinect-based virtual rehabilitation system for inpatient rehabilitation of the upper limb in patients with subacute stroke: A randomized, double-blind, sham-controlled pilot trial. Medicine. Jun 2018;97(25):e11173. [CrossRef]
    61. Pin-Barre C, Laurin J. Physical exercise as a diagnostic, rehabilitation, and preventive tool: influence on neuroplasticity and motor recovery after stroke. Neural Plast. Nov 22, 2015;2015:608581. [CrossRef] [Medline]
    62. Ribeiro de Souza F, Sales M, Rabelo Laporte L, Melo A, Manoel da Silva Ribeiro N. Body structure/function impairments and activity limitations of post-stroke that predict social participation: a systematic review. Top Stroke Rehabil. Sep 05, 2023;30(6):589-602. [CrossRef] [Medline]
    63. Chen L, Chen Y, Fu W, Huang D, Lo W. The effect of virtual reality on motor anticipation and hand function in patients with subacute stroke: a randomized trial on movement-related potential. Neural Plast. Jan 24, 2022;2022:7399995. [CrossRef] [Medline]
    64. Perez-Marcos D, Bieler-Aeschlimann M, Serino A. Virtual reality as a vehicle to empower motor-cognitive neurorehabilitation. Front Psychol. Nov 2, 2018;9:2120. [FREE Full text] [CrossRef] [Medline]
    65. Cameirão MS, Faria AL, Paulino T, Alves J, Bermúdez i Badia S. The impact of positive, negative and neutral stimuli in a virtual reality cognitive-motor rehabilitation task: a pilot study with stroke patients. J NeuroEngineering Rehabil. Aug 9, 2016;13(1):70. [CrossRef]
    66. De Luca R, Lo Buono V, Leo A, Russo M, Aragona B, Leonardi S, et al. Use of virtual reality in improving poststroke neglect: promising neuropsychological and neurophysiological findings from a case study. Appl Neuropsychol Adult. Sep 22, 2019;26(1):96-100. [CrossRef] [Medline]
    67. Johansson B. Current trends in stroke rehabilitation. A review with focus on brain plasticity. Acta Neurol Scand. Mar 2011;123(3):147-159. [CrossRef] [Medline]
    68. Ekman U, Fordell H, Eriksson J, Lenfeldt N, Wåhlin A, Eklund A, et al. Increase of frontal neuronal activity in chronic neglect after training in virtual reality. Acta Neurol Scand. May 16, 2018;138(4):284-292. [CrossRef]
    69. Prochnow D, Bermúdez i Badia S, Schmidt J, Duff A, Brunheim S, Kleiser R, et al. A functional magnetic resonance imaging study of visuomotor processing in a virtual reality-based paradigm: rehabilitation gaming system. Eur J Neurosci. May 17, 2013;37(9):1441-1447. [CrossRef] [Medline]
    70. McDermott EJ, Metsomaa J, Belardinelli P, Grosse-Wentrup M, Ziemann U, Zrenner C. Predicting motor behavior: an efficient EEG signal processing pipeline to detect brain states with potential therapeutic relevance for VR-based neurorehabilitation. Virtual Real. Sep 23, 2023;27(1):347-369. [CrossRef] [Medline]
    71. García-Murillo DG, Alvarez-Meza A, Castellanos-Dominguez G. Single-trial kernel-based functional connectivity for enhanced feature extraction in motor-related tasks. Sensors (Basel). Apr 13, 2021;21(8):2750. [FREE Full text] [CrossRef] [Medline]
    72. Ferrazzoli D, Ortelli P, Volpe D, Cucca A, Versace V, Nardone R, et al. The ties that bind: aberrant plasticity and networks dysfunction in movement disorders-implications for rehabilitation. Brain Connect. May 01, 2021;11(4):278-296. [CrossRef] [Medline]
    73. Sui Y, Kan C, Zhu S, Zhang T, Wang J, Xu S, et al. Resting-state functional connectivity for determining outcomes in upper extremity function after stroke: a functional near-infrared spectroscopy study. Front Neurol. Nov 9, 2022;13:965856. [FREE Full text] [CrossRef] [Medline]
    74. Harirchian MH, Rezvanizadeh A, Fakhri M, Oghabian MA, Ghoreishi A, Zarei M, et al. Non-invasive brain mapping of motor-related areas of four limbs in patients with clinically isolated syndrome compared to healthy normal controls. Journal of Clinical Neuroscience. Jun 2010;17(6):736-741. [CrossRef]
    75. Wu Q, Yue Z, Ge Y, Ma D, Yin H, Zhao H, et al. Brain functional networks study of subacute stroke patients with upper limb dysfunction after comprehensive rehabilitation including BCI training. Front Neurol. Jan 27, 2019;10:1419. [FREE Full text] [CrossRef] [Medline]
    76. Alves da Cruz M, Laurino M, Christofaro D, Ghisi G, Vanderlei L. Long-term effects of virtual reality-based therapy in cardiovascular rehabilitation: a longitudinal study. Physiother Theory Pract. Apr 25, 2024;40(4):727-735. [CrossRef] [Medline]

    ‎
    ARAT: Action Research Arm Test
    BBT: Box and Block Test
    CON: conventional therapy
    FIM: Functional Independence Measure
    FMA-UE: Upper Extremity Fugl-Meyer Assessment
    MD: mean difference
    MINORS: Methodological Index for Non-Randomized Studies
    PEDro: Physiotherapy Evidence Database
    PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses
    RCT: randomized controlled trial
    VR: virtual reality
    VRT: virtual reality therapy
    WMFT: Wolf Motor Function Test

    Edited by A Mavragani; submitted 23.09.24; peer-reviewed by Z Ren, FZ da Silva Areas; comments to author 04.12.24; revised version received 31.12.24; accepted 06.01.25; published 19.02.25.

    Copyright

    ©Shiqi Xu, Yanwen Xu, Ruyi Wen, Jun Wang, Yuyu Qiu, Chetwyn CH Chan. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 19.02.2025.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.