Stroke-induced upper limb dysfunction requires functional assessment and rehabilitation. The intelligent rehabilitation assessment and virtual reality training system for upper limb motor function in stroke can accurately and objectively assess patients’ motor function and guide their rehabilitation training. Our study aims to verify the clinical efficacy of the virtual reality training system in improving upper limb motor dysfunction in poststroke patients.

This study will be a single-center, single-blind, randomized controlled clinical study. Fifty eligible patients will be randomized in a 1:1 ratio into a virtual reality training group (VR) and a conventional upper extremity treatment (CT) group. The intervention will be performed five times per week for 4 weeks. The primary outcome will be the Fugl-Meyer Motor Function Assessment—Upper Extremity (FMA-UE), and the secondary effects will be kinematic and electromyographic assessments. Adverse events will be recorded, and serious adverse events will be used as criteria for discontinuation of the intervention.

A stroke upper limb motor function assessment and virtual reality rehabilitation training system based on the FTHUE scale can achieve a close link between intelligent assessment and treatment of upper limb motor function in poststroke patients while integrating the design concepts of the upper limb and hand assessment and treatment, which can theoretically improve upper limb function in stroke to a greater extent, but further high-quality studies are needed. The results of this trial will determine whether an assessment and training system based on the FTHUE scale can improve upper extremity motor dysfunction after stroke.

Chinese Clinical Trial Registration Center, ChiCTR2200060214. Registered May 22, 2022. Manuscript Version: 2.0 Manuscript Date: May 2, 2025.

Stroke, a sudden-onset cerebrovascular disease, has become one of the leading causes of disability worldwide, and its impact on patients’ quality of life is severe^[1]. Among the many dysfunctions caused by stroke, upper limb dysfunction is particularly prominent, with more than 80% of patients experiencing contralateral upper limb and hand dysfunction. Most patients have difficulty achieving optimal recovery of upper limb function, making it difficult to perform basic movements such as dressing, eating, and writing independently. Even after rehabilitation, about 65% of patients are unable to effectively integrate their damaged upper limbs into daily activities after 6 months, and 71% of patients still had significant upper limb or hand functional limitations when performing complex tasks^[2]. Therefore, improving the functional use of the upper limbs in stroke patients is vital in enhancing their self-care ability and promoting social participation.

During stroke rehabilitation, traditional rehabilitation therapies such as constraint-induced movement therapy^[3], mirror therapy^[4], and activity-based task-oriented training^[5] have been shown to have some efficacy. Traditional therapies such as occupational therapy and physical therapy, through systematic rehabilitation training, can also help patients improve their upper limb motor deficits after stroke to some extent. However, these methods are often time-consuming and cumbersome, requiring patients to invest time and energy. In addition, post-stroke patients are in urgent need of a low-cost and motivating rehabilitation technique to meet their rehabilitation needs due to increased medical costs or lack of access to rehabilitation specialists.

The advent of virtual reality (VR) technology has brought new hope to stroke rehabilitation. It is widely used in rehabilitating many diseases such as stroke, Parkinson’s, and multiple sclerosis^[6] and has shown great potential for improving upper limb function, cognitive function, balance, and walking ability in post-stroke patients^[7]. In 2016, the Joint American Heart Association and American Stroke Association Guidelines for Adult Stroke Rehabilitation even listed VR technology as a Class IIa, Level B recommendation for use in functional rehabilitation of the upper extremity after stroke^[8]. Systematic Reviews by Laver et al. show that VR technology significantly improves upper extremity function and activities of daily living in post-stroke patients^[9]. Mekbib also noted that VR training reduces upper limb dyskinesia in stroke patients and encourages active participation in exercise and social activities^[10]. Analysis of the study showed that the longer the treatment time (more than 15 h) of the involvement in virtual reality training, the more significant the improvement in upper extremity dyskinesia and activity limitation^[11]. In addition, patients in the subacute phase are more likely to benefit from virtual reality therapies than those in the chronic phase, as brain plasticity and cortical reorganization typically reach higher levels in the days or weeks following a stroke^[12].

Although the virtual reality rehabilitation system shows excellent potential in stroke rehabilitation, there are still some problems. First, the intelligent assessment of upper extremity motor function is mainly focused on the FMA scale, Brunnstrom scale, and Wolf scale, while the FTHUE scale, widely used in clinical practice, has not yet been thoroughly investigated^[13]. Secondly, there are deficiencies in VR-based upper limb functional rehabilitation systems, such as the lack of patient-centered and task-oriented individualized training programs, less research on the upper limb and hand as a whole, more subjective clinical efficacy assessment, and the lack of close integration of evaluation and treatment^[14,15,16]. The FTHUE assessment and rehabilitation program has been widely used in clinical practice, and the FTHUE scale mainly evaluates patients’ upper limbs and hands after stroke. The FTHUE scale assesses the ability of post-stroke patients to use their upper limbs in daily life. At the same time, the FTHUE rehabilitation program is based on neurodevelopmental techniques and motor relearning theory, and patient-centered task-oriented training is beneficial for upper limb functional performance in stroke patients^[17].

Combining the FTHUE rehabilitation program with VR training can promote patients’ motor function recovery more effectively. Selecting the appropriate FTHUE rehabilitation program based on the objective assessment results of the FTHUE Assessment Scale allows for developing a personalized training plan and effectively implementing rehabilitation training principles. In addition, systems designed with high-precision, markerless sensing technologies, such as the Azure Kinect and Leap Motion devices, provide low-cost, portable solutions suitable for community or home environments.

Based on these advances, this study aims to conduct a clinical investigation to verify the feasibility, effectiveness, and safety of the developed rehabilitation training system for post-stroke patients from biomechanical and neurophysiological perspectives. This study will elucidate the specific mechanisms by which VR training improves upper limb motor function, providing theoretical support for optimizing VR technology in clinical practice and advancing the field of stroke rehabilitation. Ultimately, this work aims to provide superior rehabilitation solutions for stroke survivors.

This is a parallel, single-blind, randomized controlled trial based on a superiority framework design to determine whether performing VR training results in better outcomes in upper limb motor function in stroke patients compared to conventional rehabilitation. Result evaluators and data analysts will be blinded. Fifty eligible participants are randomly assigned to the VR group and the control group in a ratio of 1:1. The VR group receives 4 weeks of VR training five times a week, each time for 30 min, a total of 20 times. The control group does not receive VR training. Both groups receive medical treatment and routine rehabilitation training. Primary and secondary outcomes will be measured at baseline and 4 weeks (end of intervention). This study uses the SPIRIT reporting guidelines^[18]. The flow chart of the research design is shown in Fig. 1, and the time point of the research visit is shown in Fig. 2. An additional file shows this in more detail (see Additional file 3 SPIRIT Figure & Flow diagram.docx).

Flow diagram of participants. Abbreviations: FMA-UE, the Fugl-Meyer assessment of the upper extremity; MAS, Modified Ashworth Scale; WMFT, Wolf motor function test; SIS, Stroke Impact Scale; Kinect, Azure Kinect kinematic analysis; sEMG, surface electromyography

Full size image

SPIRIT figure. Schedule of enrolment, interventions, and assessments

Full size image

Participants will be all from the Rehabilitation Hospital of Fujian University of Traditional Chinese Medicine in Fuzhou, China. Data collection will divide into two phases: the first phase is a baseline test completed within 1 week before the intervention is implemented; the second phase is a final test conducted within 1 week after the intervention is completed.

To be eligible to participate in the study, participants must meet the following required criteria: (1) meeting the diagnostic criteria for “stroke” established at the fourth national cerebrovascular disease academic conference in 1995 and confirmed by head CT or MRI; [2] first-time stroke with a disease course ranging from 2 weeks to 6 months; [3] age between 20 and 70 years; [4] blood pressure stably below 130/100 mmHg; [5] Mini-Mental State Examination (MMSE) scores: illiterate individuals > 17 points, primary school education > 20 points, and middle school education or above > 24 points; [6] affected upper limb at Brunnstrom stage III or above; [7] willingness to sign the informed consent form and understand, accept, and follow the rehabilitation guidance.

The exclusion criteria for participants are as follows: [1] upper limb motor dysfunction attributed to other brain diseases, including brain tumors, traumatic brain injury, or parasitic brain disease; [2] presence of conditions that may impair training efficacy, such as arthritis, joint injuries; [3] patients with severe post-stroke complications, such as severe pulmonary infections, shoulder-hand syndrome; [4] Individuals with severe cardiovascular disease, heart, liver, or kidney failure, malignancies, or gastrointestinal bleeding; [5] severe visual impairment that precludes the completion of the training; [6] skin or muscle lesions of the extremities that interfere with surface electromyography; [7] inability to cooperate or participate in the study’s examination, evaluation, and treatment due to other reasons, such as intolerable pain; [8] participation in other concurrent clinical studies.

Informed consent for this study will be obtained by GCP-certified research professionals (QZ and JL, licensed rehabilitation therapists). Specific procedures included the distribution of a study briefing booklet after the initial screening in the outpatient clinic, followed by a standardized 30-min informed communication in a separate consultation room. It will be formally signed using a dual signature system (original on file and copy to the subject). For special populations such as illiterates, this study required the participation of a notary witness throughout the entire process. All consent processes will be audio-visualized and retained, and the paper originals will be kept under a standardized double-lock for 5 years.

The informed consent form for this study will inform participants of the assessments involved and emphasize that all data collected will be limited to the use of this study and will not be reused outside of the study. In addition, this study will not involve biological samples from the participants, so there will be no need to obtain consent from the participants again.

Constraint-induced movement therapy (CIMT), mirror therapy, and task-oriented training will be selected as controls in this study, mainly based on the following scientific rationale: first, all of these methods are classified as Class A recommended evidence (Level of Evidence Ia–Ib) by the Chinese Guidelines for Stroke Rehabilitation Treatment (2021) and AHA/ASA Guidelines (2016), which represent the three primary evidence-based intervention in stroke rehabilitation, respectively paradigms (Compulsory Use Theory, Visual-Motor Integration Mechanisms, and Function-Specific Reorganization Theory); and second, all control interventions will be strictly matched for treatment dosage (60 min/treatment, five treatments/week) and primary evaluation metrics (Fugl-Meyer Upper Extremity Score). The study will use a superiority design to mimic real clinical practice by setting up a “combination therapy” control group (CIMT + mirroring + task training) to ensure that patients in the control group receive the best treatment available and that risks are minimized.

Participants in this cohort will receive standard care and rehabilitation protocols. They will receive conventional therapies such as etiology-based treatments, blood pressure regulation, antiplatelet and anticoagulant therapies, cardiovascular management, hyperglycemia control, and glucose and lipid management. Certificated occupational therapists will perform rehabilitation therapy focusing on hand and arm movements. Each subject will participate in 20 1-h sessions over 4 weeks, five times per week. Throughout the intervention, all subjects will maintain their diet and medication intake records.

The treatment group will conduct VR training based on routine medical and rehabilitation training. Basic treatments are the same as those in the control group. The VR training and rehabilitation are 30 min each time, once a day, five times a week for 4 weeks, respectively, and are conducted by certified occupational therapists.

Participants will withdraw from the trial according to the following conditions: [1] participants experiencing severe adverse events and deemed unsuitable for the test; [2] subjects who experience a recurrent stroke after enrollment; [3] subjects who voluntarily chose to withdraw from the trial; [4] patients in this study requiring emergency interventions due to worsening conditions or severe complications.

During the training process, we will adopt a gamification design to enhance patient participation and enthusiasm through enjoyable and interactive VR games; develop personalized training programs, adjusting the content and intensity according to the patient’s rehabilitation stage and ability level; provide timely feedback, monitoring performance in real-time during training and giving intuitive feedback and rewards; enrich training scenes and content, creating diverse tasks with the help of VR technology to avoid monotony; increase social interaction, designing a multi-participation mode for patients to train together with their relatives, friends or patients; optimize the ease of use of the system to ensure operation and friendly interface; and professional guidance and support. Social interaction, designing a multi-participant mode for patients to train with their relatives or friends; optimizing the system’s ease of use to ensure simple operation and a user-friendly interface; and professional guidance and support, with a rehabilitation therapist to provide professional help throughout the process. The combined use of these strategies can significantly improve the compliance of stroke patients in VR rehabilitation training and thus enhance the rehabilitation effect. A series of measures will be taken to improve the quality of the trial. Participants in the virtual reality group will be trained at a specified time and place. In addition, two research assistants will closely monitor the VR training process to ensure that participants are fully engaged.

In principle, participants in the intervention and control groups will receive no additional rehabilitation training during the intervention period. Participants’ attendance will also be ensured.

During the study intervention, any adverse events experienced by participants will be monitored and documented in the adverse event case report form and then reported to the research assistant. The potential causal relationship between VR training and the severity of these adverse events will be analyzed.

Any adverse events, defined as any dysfunction caused by the intervention (e.g., shoulder pain), will be recorded on the case report form (CRF). In an adverse event, the coach or project manager will provide additional treatment for the participants. Serious adverse events must be reported to the ethics committee of the rehabilitation hospital affiliated with Fujian University of Traditional Chinese Medicine.

Based on various factors, the safety of participants will be evaluated using the following criteria: Level 1 (safe, no adverse reactions); Level 2 (relatively safe, with mild adverse reactions that do not require treatment and allow participants to continue training); Level 3 (moderate adverse reactions that require treatment but allow participants to continue training); and Level 4 (study termination due to severe adverse reactions). These levels are based on the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0, US Department of Health and Human Services, 2017.

This trial’s variables encompass basic characteristics, primary and secondary outcomes, and exploratory results. Basic characteristics will be assessed via a questionnaire at baseline (1–2 weeks before randomization). Primary and secondary outcomes and exploratory results will be evaluated at baseline and the intervention’s conclusion (4 weeks post-randomization). All assessments will be conducted by experienced evaluators blinded to participant assignments.

Primary outcome measures

The Fugl Meyer upper limb assessment scale (FMA-UE).

The FMA-UE comprises 33 items, segmented into four subscales: arm, wrist, hand, and coordination, with scoring based on a 3-level sequential scale^[20]. Scores reflect the ability to perform independent actions within and outside synergy patterns. A maximum score of 66 indicates normal motor function. The FMA-UE is widely used to assess stroke severity, demonstrating strong reliability and validity. Additionally, the FMA-UE evaluates sensation (0–12), passive range of motion, and pain (0–24).

Secondary outcome measures

The kinematic analysis.

The kinematic analysis encompasses endpoint performance metrics, joint angles, shoulder/elbow coordination indices, and compensatory trunk movements. Two Kinect motion capture systems are employed to evaluate the three-dimensional motion of the trunk and upper limbs during a reaching-to-grasp-pen task. The high-resolution camera units calculate the 3D coordinates of markers in real time, with data automatically collected via Kinect SDK and subsequently transferred to Matlab for customized analysis. The Azure Kinect records XYZ coordinates of 25 anatomical landmarks at 30 Hz without requiring physical markers. These landmarks include the head, neck, shoulders, spine, elbows, wrists, hands, fingertips, thumbs, hips, knees, ankles, and feet. The Azure Kinect is mounted on a 114-cm tripod at a 20-degree angle to ensure stable recording, positioned 140 cm from the table’s far end. The table and chair are 75 cm and 46 cm tall, respectively. Participants sit comfortably in front of the Kinect, naturally performing the reaching-to-grasp-pen task at their own pace. A video guides them and repeats each task five times without additional assistance, except for safety concerns.

The Modified Ashworth Scale (MAS).

The Modified Ashworth Scale (MAS) assesses increases in passive motor resistance and spasticity^[21]. A 6-point sequential scale measures muscle tension in the shoulder adductor, elbow, wrist, and finger flexors at rest, with scores ranging from 0 (no increase in muscle tension) to 4 (extreme rigidity during flexion or extension). A MAS score of ≥ 1 + in one or more muscles signifies the presence of a spasm.

Wolf motor function test (WMFT).

The WMFT scale assesses upper limb functional activity^[22], comprising 15 tasks: items 1–6 involve basic joint movements, while items 7–15 focus on complex functional activities. Scores are assigned based on performance quality, ranging from 0 to 5 points across six levels.

The Stroke Impact Scale (SIS).

The Stroke Impact Scale (SIS) evaluates the overall quality of life in stroke survivors^[23]. It comprises self-reported questions assessing various domains, including strength, hand function, mobility, and daily activities, yielding a total score from 16 to 80 (severe impairment) to 80 (minimal impairment).

Virtual Reality Rehabilitation Training System Questionnaire.

A survey is administered to evaluate the subjects’ usability and satisfaction with the developed system. The rating scale ranges from 1 (“completely disagree”) to 5 (“completely agree”). The evaluation covers the following aspects: the difficulty of the training, the interactivity of the game, the ease of operation, whether it is more enjoyable compared to traditional training modes, and whether it enhances the motivation of the training.

Exploratory outcomes measures

Electromyogram data acquisition.

EMG signals are captured from the bilateral superior trapezius, anterior deltoid, posterior deltoid, middle deltoid, biceps brachii, triceps brachii, extensor carpi radialis, and flexor carpi radialis. Surface EMG is recorded at 1500 Hz using the Noraxon system with disposable Ag–AgCl electrodes in a bipolar configuration, with an inter-electrode distance of 2 cm. Following standard skin preparation, electrode pairs are positioned over the muscle bellies, identified by palpation and brief elbow flexion–extension tests. EMG signals are monitored in real-time before the experiment to ensure proper electrode placement and recording quality.

These assessments will be conducted before and after a 4-week intervention by researchers who are not involved and are blinded to the group assignments.

The study will assess primary, secondary, and exploratory outcome indicators and descriptive data for both phases. Participants will be assessed at baseline and the end of the intervention. We will measure participants’ upper extremity motor function, ability to perform activities of daily living, and muscle synergy patterns through the collection of data (e.g., clinical scales, and surface electromyography). Test results will be recorded in a timely manner to ensure comprehensive data collection. Program testers will remain consistent throughout the process. To minimize participant attrition, researchers will inform subjects of measurement details before the start of the assessment to facilitate communication with subjects, promote communication, and reduce participant attrition. The study schedule is detailed in Fig. 1.

This study observes VR’s effect on poststroke patients’ upper limb function. When estimating the sample size, the FTHUE scale is the primary impact indicator to calculate the required sample size. According to relevant published literature^[19], the FTHUE scores of poststroke patients in the VR and control groups after intervention are 5.00 ± 1.2 and 4.2 ± 0.78 points, respectively. According to the above results, a sample size of 42 participants is calculated to sufficiently detect the target effect size (0.7904922) with a type 1 error of 5% (α = 0.05) and 80% power (β = 0.20) by Gpower V.3.1.9.2 software. Assuming a dropout rate of 20%, a total of 50 participants is necessary, with 50 participants in each group.

A total of 50 participants are recruited from the rehabilitation hospital affiliated with the Fujian University of Traditional Chinese Medicine in Fuzhou, China. Participants will be recruited by distributing leaflets, posters, and recruitment information on WeChat or other online platforms. Interested volunteers can contact the research assistant, who will screen applicants according to the inclusion and exclusion criteria. The research assistant introduces the fundamental research and informed consent to the volunteers, including the research cycle and intervention methods. At the same time, the research assistant explains the grouping situation. Participants should obey the random assignment and follow the scheme specified by each group (if randomly assigned to the control group, participants must agree not to participate in VR training); otherwise, they are excluded. Eligible volunteers are invited to join the study, sign the informed consent form, and then arrange the baseline assessment.

After baseline assessment, eligible participants are randomly assigned to the VR group and the control group in a ratio of 1:1. Random assignment sequences are generated by independent statisticians using the planned program of statistical software SAS 9.1.

The randomization sequence is hidden by opaque and sealed letters, and participants are managed by an independent research assistant who knows nothing about recruitment, evaluation, and intervention. The independent research assistant informs eligible participants of their assignment results by telephone.

To prevent contamination of the intervention, participants will be assigned to either the intervention or the control group using a strict simple randomization method that informs participants of the group to which they belong and the intervention to which they should adhere and restricts the control group from receiving additional interventions. Educate and train participants, emphasizing the importance of adherence to the study protocol while ensuring compliance through regular monitoring and assessment, such as questionnaires, interviews, or use of logbooks, and timely detection and correction of deviations from the protocol.

In this study, blinding participants is not feasible. However, we will implement blinding for statisticians and evaluators.

A two-tier blinding approach is used: initially, participants are assigned to groups represented by the letters “A” and “B.” In the second tier, “A” and “B” are used to denote the intervention types, such as “VR Group” and “Control Group.” Data will be entered into the database after the study is completed using participant codes. Following data analysis, the database will be locked for primary unblinding to reveal which participants correspond to “A” and “B.” Final secondary unblinding will determine whether participants received the VR or control intervention.

Comprehensive data collection and recording will be performed on the experimental and control groups. This will include an assessment of upper limb motor function, an evaluation of activities of daily living (ADLs), a kinematic assessment, and surface EMG measurements. Assessments will be performed before and 4 weeks after the intervention. At the time of the evaluation, subjects will receive an informed consent form. Subjects will be asked to sign the informed consent form. Testing will be performed by members of the research team who do not have a conflict of interest, and these testers will not have access to detailed information about the assignment of tasks to the intervention group.

Before initiating the project, a comprehensive research protocol manual and a standardized case report form are meticulously developed. To ensure the consistency and accuracy of the research procedures, neuropsychological experts are engaged to conduct detailed training sessions for all evaluators involved. This training aims to familiarize them with every aspect of the project implementation process, guaranteeing the study’s high quality.

We will set up a follow-up management team consisting of a study coordinator, a clinical follow-up specialist, and a data manager, develop a standardized operation manual, and implement the “3–2-1” contact strategy (send message 3 days in advance → phone call 1 day before → confirmation on the same day). Multiple contact information will be collected, and an electronic reminder system will be set up during the baseline period. During the follow-up period, a time window of ± 7 days will be provided, and multiple follow-up methods will be available (outpatient/telephone/door-to-door), with transportation subsidies and incentives set up. We will clarify the criteria for lost visits (3 consecutive contact failures) and initiate a graded remediation process (therapist home visit → PI call → community assistance). A quality control committee consisting of a project professor, two associate professors, and PhD members will review the follow-up completion rate (target ≥ 90%), data completeness, and protocol deviations monthly, implementing three-tiered quality control of investigator self-inspection-coordinator weekly verification-committee monthly inspection.

Data will be collected in this study using paper case report forms (p-CRF), with data quality monitored and entered into Microsoft Office Excel by a research assistant and subsequently imported into a data management platform managed by an independent institution (FJTCM Yun, http://10.252.47.2). All paper forms will be kept in institutional locked dedicated filing cabinets (access controlled) accessible only to core members (PI and Data Security Specialist) and retained electronically for 6 months, i.e., shredding and destruction will be supervised by the Specialist with signed records. We will implement stringent data management measures: dual independent entry (a third person reviews discrepancies); electronic data is stored on the organization's encrypted server (weekly backup) and automatically erased after 5 years of retention; version control is activated to record all modifications, and 10% of the data is randomly sampled by the commissioner for manual review. Unique identifiers will be assigned to each participant, consent forms will be stored encrypted separately from data (only the PI and commissioner held keys), and pseudonyms will be used in reports. Junior researchers require panel approval to access the data, and data security specialists supervise the whole process but do not have access to detailed information. The study will follow the approval of the ethics committee and the Declaration of Helsinki. Participants who withdraw from the trial will be given the option to provide follow-up data. Upon completion of the study, scanned copies of the original forms (de-identified) are available upon request, and other results will be shared through academic papers and trial registration platforms.

Neither the data sheets nor the electronic data files collected will contain personal information. Each participant will be given a unique study code to ensure data anonymity.

Data collection in this study do not involve biological specimens from participants.

The research data will be analyzed using IBM SPSS Statistics 20.0. Measurement data are presented as mean ± standard deviation, while counting data are described in frequency, percentage, or proportion. For normally distributed samples, paired sample t-tests are employed for within-group comparisons, and independent sample t-tests are used for between-group analyses. The Wilcoxon signed-rank test is applied to samples that did not follow a normal distribution. All statistical tests are two-sided with a significance level of α = 0.05. A P value of ≤ 0.05 is considered to indicate statistical significance.

we will use a dual comparison strategy: the primary analysis will compare the absolute scores of the two groups at the point of the end of the 4-week intervention (T1) (cross-sectional comparison), and the secondary analysis will compare the value of the change in each group relative to baseline (longitudinal comparison). Specifically, [1] for continuous variables (FMA-UE scores, MBI index, sEMG synergy index), mean ± standard deviation (normally distributed data) or median [IQR] (non-normal data) will be reported at the same time, and between-group comparisons will be made using independent samples t-tests or Mann–Whitney U-tests; and [2] additional calculations of clinical response rates (dichotomous: e.g., FMA-UE improvement of ≥ 9 points is defined as effective), and between-group differences will be analyzed using chi-square tests; [3] all longitudinal change values will be supplemented with effect sizes (Cohen’s d) reported. This composite analysis method assesses both the immediate intervention effect and the degree of individual progress.

Behavioral indicators of upper limb function level

Independent statisticians who are not involved in result evaluation will analyze all data using the SPSS V.24.0 software package, with statistical significance set at a two-sided P value of less than 0.05. Normally distributed continuous variables are presented as mean ± standard deviation (SD), while non-normally distributed variables are described by the median and interquartile range (IQR). Categorical variables are reported as frequencies or percentages.

The study will collect basic data, including age, gender, and BMI, and include these as covariates in the analysis. Differences between the VR and control groups at each time point (4 weeks post-intervention) are analyzed using Student t-tests or Mann–Whitney U tests. A linear mixed model with restricted maximum likelihood is employed to explore group × time interactions. Correlation analysis is conducted using Pearson or Spearman correlation coefficients.

Kinematic data analysis

The University of Western Australia (UWA) upper limb model is utilized for three-dimensional motion analysis within the landmark motion analysis system, incorporating 18 markers. The trunk, upper arm, forearm, and hand segments are delineated based on anatomical marker locations. Calibration Anatomical System Technology facilitates the establishment of markers relative to the upper arm group (PUA) or forearm group (DUA) coordinate systems. Upper limb landmark movements are reconstructed relative to the upper limb technical coordinate system. During tasks, the Kinect system records the three-dimensional coordinates of anatomical markers recognized from its skeletal model. Local segment coordinates, including those for the torso and upper arm, are established based on global coordinates.

The Azure Kinect system utilizes Kinect SDK2.0 and Microsoft Visual Studio 2016 (Microsoft Cop., Redmond, WA, USA) to capture and save kinematic data and the time range. Matlab v2018a (MathWorks Inc.; MA, USA) is employed to import and process the 3D position data from the Kinect sensor. The Azure Kinect captures XYZ coordinate data for 25 body joints, with kinematic data from ten specific joints being utilized, including the spine, shoulders (left/right), elbows (left/right), wrists (left/right), and hands (left/right). Each repetition’s data is segmented and filtered using singular spectrum analysis (SSA) to mitigate noise. The team’s previously developed method calculates kinematic indices from the filtered data. For the Kinect upper limb evaluation system, the spatial positions of measured nodes serve as inputs, with the developed markerless motion analysis system computing joint angles and spatiotemporal parameters. A Butterworth low-pass filter with a 6-Hz cutoff frequency is applied. The customized upper limb kinematics model calculates the three Euler angles of shoulder rotation for the Azure Kinect system, following the sequence of flexion (+)/extension (−), adduction (+)/abduction (−), and internal rotation (+)/external rotation (−). Elbow flexion is determined using trigonometric functions based on the position data of ShoulderRight, ElbowRight, and WristRight. Matlab 2018a is used to develop the kinematics model for the Azure Kinect system.

Muscle data analysis

In this study, we will use a muscle synergy analysis method based on non-negative matrix factorization to investigate the coordination patterns of muscles during exercise. First, the acquired multichannel surface electromyographic signals are preprocessed, including filtering, noise reduction, and normalization operations, to improve signal quality and reduce noise interference. Subsequently, the preprocessed surface electromyographic signal matrix is decomposed into muscle synergy pattern matrix (W) and activation coefficient matrix (H) using the NMF algorithm to extract potential muscle synergy patterns. In addition, to further analyze the dynamic properties of muscle synergy, we also incorporate time–frequency analysis methods, such as wavelet transform, to capture the characteristic performance of muscle synergy in different frequency bands.

The study will use a pre-determined interim analysis design, performed by an independent statistical team under strictly blinded conditions (using coded identifiers and without blinding) at 2 weeks post-intervention. The purpose of this interim analysis is twofold: firstly, to monitor safety and immediately initiate study suspension procedures if serious adverse events related to the protocol are detected and to carry out a termination assessment based on the available data; and at the same time, to implement adaptive adjustments that will provide a basis for optimization of the subsequent phases of the trial through the analysis of the interim results, including the possibility of scientific adjustments to the dosage regimen for the cohort of subjects to be studied in subsequent phases of the trial.

Primary, secondary, and exploratory outcomes will be analyzed by grouping subjects according to their upper extremity functional status.

This study will use the modified intention to treat analysis (mITT) as the primary analysis set, including all randomized participants who have received at least one intervention and completed baseline assessment. Multiple imputations will be used to handle missing data. Due to the particularity of rehabilitation intervention (requiring active participation of subjects), the three missing subjects who did not receive any intervention could not provide any evaluation data. Therefore, strict ITT analysis is not feasible. As a supplement, we will also report: [1] baseline feature comparison of the full analysis set (FAS); [2] sensitivity analysis following the protocol set (PPS, excluding those who have been absent continuously for ≥ 3 times); [3] record in detail the reasons for all excluded cases.

Plans to give access to the full protocol, participant-level data, and statistical code {31c}.

To safeguard participants’ privacy, the data generated or analyzed in this study will remain confidential and not publicly disclosed. However, the data can be accessed by contacting the corresponding author upon reasonable request.

Two levels of management will be set up for this study: [1] the implementation team (coordinating center) consists of data collectors and researchers who are specifically responsible for participant recruitment, data collection, and periodic intervention implementation; and [2] the trial steering committee consists of a team of assistant investigators led by the principal investigator (PI), who are mainly responsible for the study protocol development, trial coordination, and organization. In addition, we will set up a clinical trial quality control committee consisting of one professor, two associate professors, and one doctoral-level member of the project team, which will be responsible for the overall supervision of the study quality and to which the investigators will be required to submit monthly reports on the progress of the trial for review and evaluation. This three-tier management structure (implementation team–steering committee–quality control committee) will ensure the scientific execution of the study protocol and quality control throughout the entire process.

To ensure the objectivity of the study, an independent data monitoring team will be formed, consisting of disinterested third-party experts. The team will regularly review the indicators of the study participants and provide professional guidance.

The independent Data Monitoring Committee (DMC) established for this study consists of three senior experts, including a professor of neurorehabilitation (as chair), a biostatistician, and an expert in clinical research methodology. All members have more than 10 years of research experience in the relevant field, have no conflict of interest with the study, and have signed a declaration of conflict of interest. The DMC will meet quarterly for routine assessments, conduct interim analyses at 50% enrollment, and convene an emergency meeting within 48 h of a serious adverse event (SAE). The committee operates using a standardized process, receiving blinded data reports 2 weeks before the meeting, evaluating them according to preset statistical rules, and forming recommendations by majority vote. The DMC's primary responsibilities cover safety monitoring, effectiveness evaluation, and data quality control. For safety monitoring, the DMC is responsible for reviewing SAE reports, evaluating the relevance of events, and recommending study continuation/modification/termination. For validity assessment, the DMC oversees the collection of endpoint indicators, analyzes interim results, and determines whether early termination of the study is warranted. For data quality control, the committee ensured that data integrity was greater than 95%, control protocol deviations were less than 5% and oversaw randomization. The committee has the authority to make recommendations for suspension of the study directly to the Ethics Committee, to access the original case reports, and to request additional analyses. The final study report will contain the DMC conclusions. For quality assurance, a written charter was developed, all meeting records were kept for 10 years, and the effectiveness of the committee was regularly evaluated.

This study adopts a multilevel adverse event monitoring and treatment mechanism. Through the establishment of a WeChat group chat, participants can report special situations to the researcher at any time, and the researcher will immediately intervene to deal with them. Considering that although the VR intervention is low-risk, based on the characteristics of the rehabilitation patient population, we predict the possibility of transient dizziness, eyestrain, and other physiological discomforts (assessed by a 0–10-point visual analog rating scale after each intervention) as well as anxiety triggered by the immersive experience (screened by open-ended questioning at weekly follow-up visits). To comprehensively monitor unintentional injuries, we add three additional measures: [1] return visits at weeks 1, 2, 3, and 4 to collect non-systematic reports; [2] recording of device use logs to analyze abnormal patterns; and [3] blinded assessment of serious events by an independent Data Monitoring Committee (DMC). If a serious adverse event (SAE) occurs, the intervention will be immediately suspended and reported to the Ethics Committee within 24 h. Causality will be jointly determined by the PI, the Ethics Committee, and the DMC according to the WHO-UMC criteria, and it will be reported in full in the dissertation as required by the CONSORT guidelines. If SAE leads to termination of the intervention, an interim analysis will be performed based on the available data. All monitoring data and outcomes will be recorded in detail and retained for review.

The study will standardize coding of all adverse events using the MedDRA (v25.0) terminology set, with coding done independently by two professionally trained researchers and causality assessed via the WHO-UMC scale, and all data will be stored on a third-party management platform to ensure normality. We strictly follow the CONSORT harms extended statement and commit to reporting all recorded adverse events (regardless of severity) while analyzing common events with ≥ 5% incidence and all SAEs and adverse events leading to dropout in the outcome section. Although the full range of events will be reported, the analysis will differentiate between expected events (based on VR characteristic presets) and unanticipated events (as per FDA standards) and will be severity-graded using CTCAE v5.0. We implement double verification, monthly monitoring, and blinded DMC audits to ensure data quality. Statistical analyses will be performed using descriptive methods to report incidence rates, Fisher’s exact test for between-group comparisons, and separate tabular analyses for SAEs.

The research team will conduct regular field visits to the hospitals to track the effectiveness of the interventions and submit progress reports to the Trial Management Group and the coordinating organization for timely evaluation of the effectiveness of the project.

Any modifications to the protocol will be submitted by the principal investigator and approved by the Ethics Committee of Rehabilitation Hospital, Fujian University of Traditional Chinese Medicine. Upon approval, we will inform the relevant parties of the changes through the following ways: the study team will synchronize the updated protocol and ethical approvals through internal meetings or written documents; if the modification involves the informed consent form or the study process, the participants will be notified by phone, email or face-to-face and will be allowed to sign the new version of the informed consent form; the collaborating institutions will be notified by a formal letter or a revised appendix; the revisions will be synchronized on the Clinical Trials platforms (e.g., ClinicalTrials.gov) are synchronized and updated with the revisions. All notification records and participant acknowledgement documents will be kept on file.

The findings of this study will be disseminated through publication in scholarly, peer-reviewed journals and will be communicated to participants, healthcare professionals, and other relevant stakeholders. Additionally, all researchers and collaborators who contribute to this study will be acknowledged as co-authors based on the extent and significance of their contributions.

Stroke is one of the leading causes of disability worldwide, and the upper limb dysfunction it causes has a profound impact on the quality of life of patients and their families^[24]. Although there are no specific drugs to treat upper limb dysfunction, and there is a relative paucity of high-quality rehabilitation evidence, rehabilitation training has been shown to be effective in improving upper limb function. However, traditional rehabilitation training is characterized by high costs and poor patient compliance, and a safe, effective, and low-cost rehabilitation method is urgently needed.

Virtual reality (VR) rehabilitation has shown great potential in stroke rehabilitation due to its unique advantages, such as providing repetitive task training, simulating real-life scenarios, and enhancing patients'motivation to participate. Maier's proposed principles of motor skill learning emphasize the importance of key factors such as task-oriented practice, explicit and implicit feedback, and increased difficulty in motor learning and neurorehabilitation, and VR rehabilitation combines these principles with technology to effectively facilitate the recovery of upper extremity function in stroke patients^[25]. fMRI studies have also shown that neural functions of the brain are reorganized during VR training, further confirming its neurorehabilitation mechanism^[26].

Although various VR rehabilitation systems for upper limb function in stroke have been developed at home and abroad, there are still some shortcomings. For example, there is a lack of patient-centered personalized training plans, less often the hand and arm as a whole research object, the evaluation of dysfunction and clinical efficacy relies on subjective judgment, and the combination of assessment and treatment is not close enough. In addition, the current intelligent evaluation of upper limb motor function in stroke mainly focuses on the FMA, Brunnstrom, and Wolf scales, while the FTHUE scale, which is widely used in clinical practice, has not yet been adequately investigated. The FTHUE scale and its rehabilitation program have unique advantages in evaluating and improving the function of the upper limb in post-stroke patients, and combined with VR training, it is expected to achieve a more effective rehabilitation effect.

Through clinical validation, this study aims to evaluate the effectiveness of the intelligent assessment system of upper limb motor function based on the FTHUE assessment scale and the virtual reality rehabilitation training system in post-stroke patients. The study will introduce biomechanical and electrophysiological parameters as evaluation indexes, such as average joint movement speed, movement trajectory smoothness, movement coordination, and muscle activation, to further quantify the intervention effect of rehabilitation training at an objective level.

A series of measures have been implemented to enhance the quality of this trial. Participants in the VR group will conduct their training at a specified time and location. Additionally, two research assistants will closely monitor the VR training process to ensure that the participants are fully engaged in the training. However, there are some limitations to this study. First, as a non-blinded trial, participants could not be blinded, which may affect the objectivity of the results. Second, the lack of follow-up prevented assessment of the long-term maintenance of rehabilitation effects. Finally, the samples are all from the same city, limiting the generalizability of the findings. Future studies may expand the sample size, conduct multicenter studies, and conduct long-term follow-ups to verify the system’s broad applicability and lasting effects.

In conclusion, this study is the first randomized controlled trial to systematically assess the effects of VR training based on the FTHUE rehabilitation program on upper limb function in post-stroke patients from biomechanical and neurophysiological perspectives. If the trial achieves significant results, it will provide rigorous evidence and a theoretical basis for the application of VR training in patients with upper limb dysfunction after stroke, which will help to deeply understand the mechanisms of brain adaptive reorganization and neuromuscular motor control and provide valuable references for the clinical application of virtual reality technology and related research.

During the research initiation phase, the initiator will first be notified of the project, followed by the formal confirmation of the research center qualification by the project leader (PI). The revised research proposal will be submitted to the PI promptly and updated to the research platform documentation system. In case of deviation from the protocol during the execution process, the research team must use a standardized report to document the situation thoroughly and, at the same time, quickly start the protocol revision process and update the trial registration information.

FMA-UE:

The Fugl-Meyer assessment of the upper extremity

MAS:

Modified Ashworth Scale

WMFT:

Wolf motor function test

SIS:

Stroke Impact Scale

Kinect:

Azure Kinect kinematic analysis

sEMG:

Surface electromyography

Download references

Authorship will be determined in strict accordance with ICMJE standards and our institutional norms for scholarly publication, and all authors meet three basic criteria: (1) substantial contribution to study design, data acquisition, or analysis; (2) participation in drafting or significant revision of the paper; and (3) approval of the final version. Specifically, QRX will lead the study design and the framework of the paper and write the first draft; QZ and JL will be responsible for the development of test methods and data collection; XLW will be involved in the analysis of the data and interpretation of the results; QRX and BS will assess the clinical feasibility of the VR intervention and provide technical guidance; in addition, BS will supervise the progress of the study, guide the drafting of the manuscript, and provide final review of the core academic points. The order of authorship will be determined by collective consensus based on the level of contribution of the authors, and we will continue to maintain the same authorship criteria and ordering principles for subsequent publications.

Correspondence to Qiurong Xie.

This study adheres to the ethical principles outlined in the Helsinki Declaration and has obtained approval from the Ethics Committee of the Rehabilitation Hospital Affiliated with Fujian University of Traditional Chinese Medicine (Approval No. 2022KY-006-01). Before enrollment, all participants will receive comprehensive information about the trial and must provide their consent by signing the informed consent form. Participants retain the right to withdraw from the study without penalty. Further details regarding ethical approval can be found in the supplementary document [see Additional file 1: Ethical Approval Document.docx].Further details regarding the consent form can be found in the supplementary document [see Additional file 4: Model consent form.docx].

All authors have agreed to publish this article.

The author affirms that there are no conflicts of interest to disclose.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions