The clinical efficacy of virtual reality technology based on the mirror neuron theory in upper limb rehabilitation of stroke patients: a protocol for a randomized clinical trial

Abstract

Background

While mirror neuron-based rehabilitation approaches demonstrate efficacy in post-stroke upper limb motor recovery. Crucially, whether sequential activation of sensory mirror neurons preceding motor mirror neurons enhances functional outcomes remains unsubstantiated. Furthermore, conventional protocols require auditory-controlled environments and sustained high-attentional engagement for optimal efficacy. This study proposes a novel integrated intervention incorporating somatosensory observation (SO) components into Graded Motor Imagery (GMI), augmented by virtual reality (VR) technology to enhance participant engagement and attentional allocation. This synergistic approach aims to potentiate sensorimotor cortical integration, thereby optimizing upper limb recovery trajectory and clinical outcomes in stroke patients.

Methods

Sixty patients were randomized into four experimental groups: the conventional GMI Group received standard graded motor imagery; the SO-GMI Group incorporated SO with GMI; the VR-GMI Group implemented GMI through virtual reality; and the VR-SO-GMI Group combined SO and GMI within a VR environment. All interventions followed a standardized 4-week protocol. Resting-state functional MRI (rs-fMRI) assessed neuroplastic changes at baseline and post-intervention. Upper limb functional recovery was evaluated using three validated metrics: Fugl-Meyer Assessment for Upper Extremity (FMA-UE), Action Research Arm Test (ARAT), and Modified Barthel Index (MBI), administered at treatment initiation, week 2 (mid-intervention), and week 4 (conclusion) to track therapeutic efficacy. The study flow diagram is Fig. 1.

Discussion

The purpose of this clinical trial is to observe the efficacy of increased SO in GMI on the recovery of upper limb motor function after cerebral stroke through a randomized controlled clinical trial, and to explore the clinical efficacy after implementing the above therapy using immersive VR technology, as well as to further investigate whether this research protocol can achieve the neurophysiological mechanism of “sensory-motor” linkage in the brain. This research method is widely applicable to patients with poor motor function and those with limitations in active movement. At the same time, VR technology allows for one-to-many training. At the same time, VR technology can conduct one-to-many training. This study aims to improve the efficacy of graded exercise therapy for upper limb rehabilitation in cerebral stroke and provide a new method that requires less physical effort, saves manpower, and has a wide range of applicability.

Trial registration

China Clinical Trial Registry ChiCTR2400084611. Registered on 21 May 2024.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13063-026-09557-y.

Background {6a}

Stroke is characterized by high disability rates. Evidence indicates that [1] approximately 80% of stroke survivors worldwide develop functional impairments in extremities, with over half exhibiting persistent upper limb or hand dysfunction beyond 6 months post-onset. These deficits significantly compromise activities of daily living, making post-stroke upper limb rehabilitation a critical research priority in neurological recovery.

The mirror neuron system (MNS) was discovered by Rizzolatti et al. [2], the mirror network in the human brain involves the Broca area the ventral premotor cortex, the lower part of the central precentral gyrus (PreCG), the posterior part of the inferior frontal gyrus, the supramarginal gyrus, the insula, the amygdala, and the prefrontal cortex [3]. Many scholars have established treatment methods based on mirror neuron theory, including mirror therapy (MT), action observation therapy, motor imagery (MI), and GMI. GMI is a staged and progressive rehabilitation technique that combines MI and MT in a stepwise manner, which activates the cortex in stages and reduces cortical inhibition, promoting cortical reorganization and functional recovery [4]. Some studies suggest that GMI may be a sequential treatment approach that can enhance the sensorimotor cortex [5].

It is well known that the motor and sensory cortices of the brain are adjacent and closely connected. Research has confirmed that peripheral nerve stimulation in stroke patients can modulate the excitability of primary motor cortex (M1) through existing cortical-cortical connections between primary somatosensory cortex (S1) and M1 or through direct projections from thalamic nuclei [6]. Some scholars have found that in the rubber hand experiment, merely visually capturing a rubber hand placed in an anatomically consistent manner was sufficient to trigger visual-proprioceptive integration without synchronous brushing [7], indicating that sensory observation can induce visual-proprioceptive integration. Mikkel et al. [8] demonstrated that the input of visual signals may promote the activation of S1 and lead to increased excitability of M1, resulting in changes in brain plasticity. Studies have shown that sensory observation activates bilateral PreCG, left supplementary motor area, right postcentral gyrus (PoCG), left inferior parietal gyrus, and other regions, with greater clustering activation in PreCG than in PoCG, suggesting that SO may regulate the sensorimotor network by activating the MNS [9]. It has been found that mirror neurons are both motor neurons and sensory neurons [10]. While MNS-based therapies (e.g., MT, AOT) demonstrate efficacy in post-stroke upper limb rehabilitation, critical gaps remain: (1) Sequential Activation Uncertainty: Limited evidence exists on whether sequential activation of sensory mirror neurons followed by motor mirror neurons enhances functional recovery. (2) Clinical Implementation Barriers: Current protocols require minimal auditory interference and sustained high attentional engagement, severely constraining real-world applicability.

VR technology generates immersive environments that enable real-time bidirectional interaction between limb movements and multisensory feedback [11]. Research demonstrates VR-based stroke rehabilitation significantly enhances patients’ body ownership and limb position sense, with this perceptual enhancement directly improving motor functional outcomes [12]. Destaw B. et al. [13] used functional neuroimaging to confirm that VR-augmented upper limb therapy strengthens functional connectivity in motor-associated brain regions. This reorganization reflects critical cortical remodeling in lesioned areas, showing strong correlation with recovered motor function. Simultaneous EEG recordings revealed increased alpha-band power during VR training, indicating enhanced attentional focus and optimized cognitive resource allocation [14]. Critically, VR facilitates neurorehabilitation through dual mechanisms reinforced embodiment perception (body ownership/spatial localization) [12] and augmented attentional modulation [14].

Objective {7}

Hypothesis 1: Integrating SO components into GMI creates a novel therapeutic approach that strengthens functional coherence between somatosensory and motor cortices, potentially improving upper limb motor function recovery in post-stroke patients. Hypothesis 2: Compared to conventional mirror neuron therapies requiring extreme attentional focus and auditory-controlled environments, VR training boosts therapeutic efficacy by increasing engagement through multisensory immersion, optimizing attentional resource allocation and enabling simultaneous one-to-multiple training protocols.

This study aims to investigate the clinical efficacy and neurophysiological mechanisms of the proposed interventions, specifically targeting enhanced treatment response rates and clinical outcomes in stroke upper limb rehabilitation through the integration of SO and VR technologies with GMI. Concurrently, our methodology is designed to (1) establish a reduced-exertion intervention model by minimizing physical demands and (2) ensure comprehensive applicability across all upper limb dysfunction stages (Brunstrom stages I–VI). Furthermore, the implementation of a one-to-multiple training paradigm optimizes resource allocation, creating an efficient rehabilitation pathway.

Trial design {8}

This study will be conducted in the form of a randomized controlled trial.

Methods: participants, interventions, and outcomes

Study setting {9}

This study will be conducted among stroke patients in the Rehabilitation Medicine Department of the General Hospital of Ningxia Medical University in Yinchuan City, Ningxia Hui Autonomous Region, China.

Eligibility criteria {10}

Inclusion criteria:

1. 18–80 years; willingness to participate with signed informed consent.

2. Unilateral hemiplegia from radiologically confirmed (CT/MRI) ischemic/hemorrhagic stroke meeting WHO cerebrovascular disease criteria (ICD-11 8B00-8B2Z).

3. First-ever stroke within 6 months of symptom onset.

4. Mini-Mental State Examination (MMSE) score ≥ 20 (adjusted for education level per Nasreddine standards).

5. Modified Ashworth Scale ≤ 2 for wrist/hand flexors

Exclusion criteria:

1. MMSE score < 20 (validated cutoff for mild cognitive impairment).

2. Objectively assessed neglect (e.g., Catherine Bergego Scale ≥ 1) or uncorrected visual deficits (Snellen chart < 20/50).

3. Active substance abuse/dependence per DSM-5 criteria (including alcohol use disorder ≥ moderate severity).

4. Neurocognitive Barriers: Severe aphasia, attention deficits, auditory/visual processing disorders, intellectual disability, major psychiatric conditions.

5. Bilateral Motor Dysfunction: Orthopedic/neurological impairments affecting both upper limbs (e.g., cervical myelopathy, bilateral plexopathy).

6. Medical Instability: Active epilepsy (≥2 seizures/month despite medication); Organ failure (e.g., LVEF < 35%,FEV1 < 50%,eGFR < 30 mL/min).

7. Metastatic malignancies: Acute systemic illness (e.g., sepsis, decompensated cirrhosis).

8. MRI Contraindications: Ferromagnetic implants (per ACR 2023 MRI safety guidelines); Severe claustrophobia requiring sedation.

9. Pregnancy.

Interventionist qualifications

Rehabilitation training sessions were administered by certified rehabilitation therapists with ≥3 years of specialized neurological experience.

Informed consent {26a}

Prior to enrollment, the study coordinator must disclose comprehensive clinical trial details—including research objectives, methodology, potential benefits/risks, and alternative treatments—ensuring participants fully comprehend their involvement. Written informed consent, voluntarily provided after a 24-h consideration period, constitutes an absolute prerequisite for study initiation.

Additional consent provisions for the collection and use of participant data and biological specimens {26b}

Prior to obtaining informed consent, investigators will comprehensively disclose the rehabilitation intervention protocols, standardized assessment procedures, and all data collection activities that are conducted at baseline, post-intervention (week 2 and week 4), with the institutionally approved informed consent form explicitly specifying measurement methodologies and data categories.

Interventions

Explanation for the choice of comparators {6b}

While mirror neuron-based rehabilitation approaches demonstrate efficacy in post-stroke upper limb motor recovery. Crucially, whether sequential activation of sensory mirror neurons preceding motor mirror neurons enhances functional outcomes remains unsubstantiated. Furthermore, conventional protocols require auditory-controlled environments and sustained high-attentional engagement for optimal efficacy. This study proposes a novel integrated intervention incorporating SO components into Graded Motor Imagery (GMI), augmented by virtual reality (VR) technology to enhance participant engagement and attentional allocation. This study aims to investigate the clinical efficacy and neurophysiological mechanisms of the proposed interventions, specifically targeting enhanced treatment response rates and clinical outcomes in stroke upper limb rehabilitation through the integration of SO and VR technologies with GMI. Concurrently, our methodology is designed to (1) establish a reduced-exertion intervention model by minimizing physical demands and (2) ensure comprehensive applicability across all upper limb dysfunction stages (Brunstrom stages I–VI). Furthermore, the implementation of a one-to-many training paradigm optimizes resource allocation, creating an efficient rehabilitation pathway.

Intervention description {11a}

All participants will receive standardized pharmacotherapy and evidence-based conventional rehabilitation comprising five core therapeutic modules: (1) therapeutic positioning strategies, (2) occupation-based interventions, (3)comprehensive hemiparesis management, (4) graded hand function training using task-specific approaches, and (5) physical agent modalities (e.g., neuromuscular electrical stimulation). Certified rehabilitation therapists with ≥3 years of specialized neurological experience will deliver twice-daily 40-min sessions, administered 5 days/week over 4 weeks. Rigorous pre-trial protocol standardization includes: (1) therapist competency certification through Objective Structured Clinical Examinations achieving ≥85% adherence scores, and (2) treatment fidelity monitoring throughout the intervention period.

Within an acoustically controlled environment (<35 dB ambient noise verified by Type 1 sound meter), therapists administer the three-stage protocol using standardized digital interfaces. Participants seated in ergonomic chairs view training materials on 11-inch backlit tablets at 60 cm viewing distance. The first stage implements Laterality Judgment (2 min), employs the image database (200 validated limb images at 0°/45°/90°/180° rotations), where therapists randomly present stimuli using computerized randomization algorithms, asking the patient to judge whether the unilateral limb in the image belongs to the left or right side. After completing the judgment of all images, the average response time of the patient for each image and the accuracy rate of identification are manually recorded. The second stage implements MI (9 min), defined as the cognitive simulation of action without physical execution. Utilizing multimodal stimuli (video demonstrations), therapists guide patients to mentally rehearse movements of their affected limb while observing presented content. This kinesthetic imagery phase specifically targets premotor cortex activation. The third stage employs MT (9 min), wherein patients perform actions with their unaffected limb while observing real-time mirror-reflected movements of both limbs displayed on calibrated screens. Concurrently, patients engage in synchronized motor imitation through this visual feedback loop, leveraging mirror neuron systems to stimulate bilateral cortical activation. GMI protocol is administered in 20-min daily sessions, 5 days weekly across a standardized 4-week intervention period.

The SO-GMI therapy represents a novel sensorimotor integration paradigm that incorporates SO components into Stage II of conventional GMI, thereby establishing a four-phase intervention protocol. SO entails the cognitive simulation of unimplemented sensory inputs, wherein patients attentively view videos depicting tactile brushing stimuli administered to various segments of their affected limb. The standardized protocol comprises Phase I) Laterality Judgment (2 min), Phase II) SO (6 min; viewing site-specific brushing sequences), Phase III) MI (6 min; kinesthetic rehearsal of observed movements), and Phase IV) MT (6 min; bilateral motor synchronization via mirrored feedback). Administered in 20-min sessions (excluding 2-min inter-phase transitions) 5 weekdays weekly across a 4-week intervention period, this regimen delivers precisely timed neuroplasticity-targeted exposure.

VR-based rehabilitation interventions will be administered in acoustically controlled environments (<35 dB ambient noise), with participants seated in adjustable ergonomic chairs maintaining 30° elbow flexion. Workstation height will be calibrated to position the interactive surface at umbilical level to optimize upper limb range of motion. A 32-inch 4 K display (3840 × 2160 resolution, 60 cm height-adjusted) will be positioned at 1.5 m (56° horizontal viewing angle), supplemented by the Percro HIM VR headset (Beijing Noitom Technology Ltd.). Certified occupational or physical therapists holding specialized neurorehabilitation certification will guide participants through protocol execution. Using this immersive platform, subjects will be alternatively assigned to perform either GMI or SO-GMI protocols. Sessions will be conducted in 20-min daily increments, 5 days weekly over a standardized 4-week intervention period.

Criteria for discontinuing or modifying allocated interventions {11b}

Subjects shall be withdrawn from the trial if any of the following conditions occur:

1. Safety-Related Termination

   1. Occurrence of protocol-defined Serious Adverse Events contraindicating further participation.
   2. Development of emergent comorbidities compromising intervention safety.
2. Protocol Non-Adherence

   1. Completion of <90% scheduled interventions without documented medical justification.
   2. Persistent (>2 consecutive sessions) deviation from assigned intervention protocol.
3. Subject-Initiated Discontinuation

   1. Voluntary withdrawal request for any reason.
   2. Loss to follow-up (>3 consecutive missed sessions with contact failure).

Strategies to improve adherence to interventions {11c}

All rehabilitation treatments and evaluations in this study are provided without financial burden. A dedicated clinical research coordinator will maintain end-to-end engagement with participants from enrollment to study completion, establishing therapeutic alliance through continuous rapport-building. To enhance accessibility, flexible scheduling offers optimized scheduling framework including weekends, accommodating both patient and primary caregiver availability. Key stakeholder communication involves quarterly progress updates formally delivered to designated primary caregivers.

Relevant concomitant care permitted or prohibited during the trial {11d}

Ensure that all research participants receive protocol-defined standardized care, provided by qualified healthcare professionals, throughout all research phases. To optimize psychological comfort and procedural cooperation with medical procedures and research, one designated family member may accompany during all research activities (Fig. 1).

Fig. 1.

The study flow diagram

Provisions for posttrial care {30}

The interventions implemented in this study constitute established clinical practices with well-characterized safety profiles, anticipated to confer no incremental risk beyond standard rehabilitation. All participants will receive protocol-adherent standard rehabilitative care administered according to routine clinical protocols. Should any research-related adverse events occur, investigators will promptly provide medically necessary interventions, cover medically justified treatment costs, and offer appropriate compensation for research-induced injuries per institutional guidelines.

Outcomes {12}

Primary outcome

FMA-UE was employed to assess upper limb motor function, quantifying motor control, coordination, and reflex activity following stroke. This scale is a strongly recommended [15] and validated [2] outcome measure for evaluating post-stroke motor impairment in both clinical and research contexts [16]. The assessment comprises 33 standardized tasks, each scored on a 3-point ordinal scale: 0 (cannot perform), 1 (partially performs), and 2 (fully performs). The total score ranges from 0 to 66, with higher scores indicating better motor function. All assessments were administered by trained therapists or raters adhering to standardized procedures.

FMA-UE scores (presented as mean ± standard deviation) were compared within each group at three time points: baseline, 2 weeks, and 4 weeks of treatment. The primary endpoint of the study was the score at 4 weeks. The mean difference in score changes from baseline to 4 weeks was compared between the two groups to evaluate the intervention effect.

Secondary outcome measures

The change in outcome value for each measure was calculated. For scale-based measures (ARAT, MBI), within-group results are presented as mean ± standard deviation at baseline, 2 weeks, and 4 weeks. The between-group difference in the change from baseline to the 4-week time point was compared for each outcome to determine the intervention’s effect.

1. ARAT

   1. The upper limb function was assessed using ARAT, an internationally recognized, reliable, and responsive measure of upper extremity functional performance [17]. This test evaluates the actual performance and capability of the upper limb in functional tasks, specifically the ability to handle objects of varying size, weight, and shape.
   2. A trained rater observed and scored the patient’s performance on 19 functionally graded tasks (e.g., grasping blocks of different sizes, pinching balls, performing specific arm movements).
   3. Each task is scored on a 4-point scale from 0 (cannot perform) to 3 (performs normally), yielding a total score ranging from 0 to 57. A higher score indicates better functional ability.
2. MBI

   1. MBI was used to assess the patient’s actual self-care ability in personal care (e.g., feeding, dressing, bathing) and mobility (e.g., walking, using stairs).
   2. As a reliable measure of ADL [18], it rates performance across 10 ADL items via interview or observation. A 5-level scoring system is applied, with each item carrying a different weight. Scores of 0, 5, 10, or 15 are assigned based on the level of independence, yielding a total score ranging from 0 to 100. A higher total score indicates greater independence.
3. Rs-fMRI

   1. Intrinsic brain functional connectivity.
   2. All participants underwent rs-fMRI scans on a 3.0 Tesla Ingenia MRI system (Philips Healthcare, Netherlands) with a standard 12-channel head coil and analysis pipelines (e.g., functional connectivity analysis, network-based statistical modeling) will be applied to assess functional connectivity within specific networks of interest.
   3. The between-group difference in the change of specific connectivity strength metrics within the pre-defined networks of interest from baseline to 4 weeks.

Participant timeline {13}

The participant timeline is shown in Table 1.  

Table 1.

Example template of recommended content for the schedule of enrolment, interventions, and assessments

	Study period
Time point	Enrollment	Allocation	Pre-intervention	Week 2 intervention	Week 4 intervention
Designation	-T0	Day 0 (T0)	Week 0 (T1)	Week 2 (T2)	Week 4 (T3)
Enrollment
Screening	X
Informed Consent	X
Exclusion*				X	X
Randomization		X
Interventions
GMI (n = 15)			X	X	X
SO-GMI (n = 15)			X	X	X
VR-GMI (n = 15)			X	X	X
VR-SO-GMI (n = 15)			X	X	X
Outcomes
FMA-UE			X	X	X
ARAT			X	X	X
MBI			X	X	X
rs-fMRI			X		X

*X represents the time when we will finish this experimental plan

Sample size {14}

The FMA-UE served as the primary efficacy endpoint, with anticipated scores derived from prior clinical evidence [12] indicating a mean of 48.82 ± 8.32 for the VR training group versus 43.47 ± 6.65 for the control group. Using Statistical Package for the Social Sciences (SPSS) software (v26.0; IBM Corp) with α = 0.05 and power = 0.8, the initial sample size calculation yielded 24 participants (12 per arm). Accounting for a conservative 20% attrition rate based on rehabilitation trial benchmarks, the final required sample size was determined as 15 participants per group (four groups), resulting in a total sample size of 60.

Recruitment {15}

The study cohort will be exclusively recruited from Ningxia Medical University General Hospital. Prior to enrollment, potential participants will undergo rigorous prescreening by clinicians to ascertain eligibility based strictly on predefined inclusion and exclusion criteria. For eligible individuals, investigators will provide detailed explanations of research procedures, potential risks/benefits, and alternatives to participation to both participants and their legal guardians. All questions or concerns raised will be thoroughly addressed using standardized communication protocols documented in the clinical trial records. Written informed consent will be obtained from participants or legally authorized representatives prior to any study-related procedures. Subject recruitment is scheduled for imminent commencement following institutional ethical approval, with all screening processes executed per the established evidence-based criteria.

Assignment of interventions: allocation

Sequence generation {16a}

Randomization will be performed using computer-generated random numbers created by SPSS software (v26.0; IBM Corp). Immediately following participant informed consent, an independent investigator with no prior patient contact will assign each subject a sequentially numbered code corresponding to the predetermined intervention allocation via concealed implementation. This methodology ensures allocation concealment throughout the trial.

Concealment mechanism {16b}

A designated personnel uninvolved in clinical care will exclusively manage the assignment of specific random numbers to participants. This individual performs no other trial-related roles beyond number allocation, ensuring the randomization process occurs without involvement or awareness by other parties.

Implementation {16c}

Upon confirmation of participant eligibility, unique identifiers will be assigned based on stroke subtype (ischemic/hemorrhagic) and time since onset. Blinded assessors will conduct longitudinal assessments at all predefined timepoints, with evaluation results sealed and secured until database lock. Treating therapists will deliver protocol-specific interventions strictly per assigned allocation groups.

Assignment of interventions: blinding

Who will be blinded {17a}

Due to therapeutic limitations that preclude double-blinding. An independent researcher unaffiliated with patient assessments oversees blinding and randomization procedures. Outcome assessors and statisticians remain blinded throughout the trial. Two experienced therapists administer group-specific interventions in physically separate treatment zones to prevent interaction; these therapists shall not disclose allocation details to patients, assessors, or statistical personnel. The blinded independent researcher conducts data management and final statistical analysis. Blinded assessors perform evaluations in a standardized quiet environment using only patient identifiers (name, age, randomized ID).

Procedure for unblinding if needed {17b}

Clinical emergencies requiring immediate intervention (e.g., seizure, acute cerebrovascular event) necessitate unblinding to guide critical care management. When assessors suspect compromised blinding due to protocol deviations (e.g., accidental disclosure by patients), verification and documentation must occur. For participants withdrawing from the study, attending physicians receive full disclosure of prior intervention types to facilitate continuity of standard rehabilitation therapy. Unblinding is mandated in these predefined circumstances per trial safety protocols.

Data collection and management

Plans for assessment and collection of outcomes {18a}

Data collection was conducted before treatment, at 2 weeks after treatment, and at 4 weeks after treatment. All assessors were masked to the study group allocation and underwent standardized protocol training prior to assessments to minimize bias. Each outcome measure was defined using standardized criteria, and the collected data were recorded on case report forms (CRFs). Eligible participants meeting the recruitment criteria had their general information recorded, and their participation was ensured to be based on informed consent.

Plans to promote participant retention and complete follow‑up {18b}

A dedicated study coordinator was assigned to each participant from enrollment through study completion to ensure continuity of care and build a trusting relationship. Flexible appointment slots, including weekend options, were offered to accommodate the schedules of both patients and their caregivers. At critical junctures in the research, progress updates were provided to primary caregivers.

Data management {19}

Data collection will occur at three time points: before treatment, at 2 weeks post-treatment, and at 4 weeks post-treatment. The study team will strictly adhere to the planned schedule and data recording procedures. Once the treatment cycle concludes, the original records will remain unaltered, and this prohibition of changes extends fully following completion of the CRFs collection. Completed CRFs will be promptly submitted to the hospital’s Data Monitoring Committee (DMC) for review. After DMC approval, researchers will transcribe the data into Microsoft Excel.

Confidentiality {27}

The randomization list and paper copies of the CRFs must be securely stored within the clinical research office and subsequently transferred to the DMC in a timely manner. Access to these documents must be strictly limited to authorized DMC members throughout this process. Within the CRFs, any information potentially identifying participants must be coded to prevent disclosure to individuals not directly involved in the trial. De-identified data will be meticulously entered into Microsoft Excel and stored exclusively on a password-protected and physically secured computer. Any potentially identifiable patient information, including study sequence numbers, must be rigorously excluded from all subsequent publications.

Plans for collection, laboratory evaluation, and storage of biological samples for genetic or molecular analysis in this trial/future use {33}

No biological samples will be collected as part of this study.

Statistical methods

Statistical methods for primary and secondary outcomes {20a}

Primary efficacy analysis will employ intention-to-treat linear mixed-effects models with repeated measures (baseline, 4-week, 12-week). Fugl-Meyer Assessment Upper Extremity (FMA-UE) scores serve as the primary endpoint, analyzed via analysis of covariance (ANCOVA) adjusting for baseline scores and stroke type. Between-group comparisons will use Tukey’s HSD with Bonferroni correction (α = 0.0125). Secondary outcomes (WMFT, MAL, fMRI connectivity) will undergo multivariate ANOVA. Subgroup analyses will examine Brunnstrom Stage I-III patients using moderation models. Effect sizes will be reported as partial eta-squared (η < sup > 2 </sup > < sub > p </sub >) with 95% confidence intervals. Sample size (n = 15/group) provides 80% power (β = 0.2) to detect moderate effects (f = 0.25, α = 0.05) in G-Power 3.1. Missing data will be handled via multiple imputation chained equations. Statistical significance is defined as p < 0.05 (two-tailed) using SPSS (v26.0; IBM Corp).

In the rs-fMRI analysis, laterality indices (LI) for both the bilateral M1 and bilateral S1 were calculated. The LI was computed according to the following formulae [19]:

LI-M1=(Number of activated voxels in the ipsilesional M1-Number of activated voxels in the contralesional M1)/(Number of activated voxels in the ipsilesional M1+Number of activated voxels in the contralesional M1).

LI-S1=(Number of activated voxels in the ipsilesional S1-Number of activated voxels in the contralesional S1)/(Number of activated voxels in the ipsilesional S1+Number of activated voxels in the contralesional S1) [20].

Consequently, the LI value ranges from −1 to 1. An LI of −1 indicates exclusive connectivity in the contralesional hemisphere (relative to the lesion), an LI of 1 indicates exclusive connectivity in the ipsilesional hemisphere, and an LI of 0 indicates perfectly symmetrical bilateral connectivity. Thus, the LI serves as a quantitative metric to assess the symmetry of functional connectivity patterns.

Interim analyses {21b}

No interim analysis was planned for this study due to its short intervention duration and the expectation of no significant safety concerns. However, recruitment would be promptly halted should any major safety issues arise during the study period.

Methods for additional analyses (e.g., subgroup analyses) {20b}

This study does not involve any planned ancillary analyses.

Methods in analysis to handle protocol nonadherence and any statistical methods to handle missing data {20c}

If necessary, missing data will be analyzed using multiple imputation methods, supplemented by sensitivity analyses. For partially missing regional brain data in rs-fMRI, neighborhood voxel-weighted imputation will be performed.

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

Non-sensitive metadata are publicly accessible via the project’s official website. Access to raw functional MRI (fMRI) files is restricted to academic institutions and requires a signed Data Use Agreement. The complete dataset will become available within one year after study completion or upon publication of related academic papers, whichever occurs later. Requests for neuroimaging data will be processed starting two years after project conclusion.

Oversight and monitoring

Composition of the coordinating center and trial steering committee {5d}

To ensure scientific rigor and validity in this clinical trial, a comprehensive oversight and monitoring framework was established. An independent Steering Committee comprising experts in neurorehabilitation, VR technology, and clinical research supervises study progress and ensures protocol adherence. Core monitoring measures include:

• 1. Systematic Data Collection: Regular tracking of patient outcome metrics.

• 2. Biweekly Progress Reporting: Monitoring patient outcomes, VR intervention compliance, and adverse events.

• 3. Monthly Internal Audits: Verifying procedural compliance and data accuracy.

• 4. Iterative Improvement Mechanism: Integrating feedback from healthcare providers and patients to optimize interventions.

• 5. Formal Accountability Protocols: Defined responsibilities and procedures for addressing non-compliance to maintain study integrity.

Collectively, these safeguards ensure operational transparency, clear accountability, and scientific reliability throughout the trial.

Composition of the data monitoring committee, its role and reporting structure {21a}

The primary functions of the DMC are to safeguard participant safety, ensure data integrity, monitor treatment efficacy, and provide recommendations regarding trial continuation, modification, or termination. Composed of independent experts in clinical trials, biostatistics, and research ethics, the DMC reports directly to the study sponsor or Steering Committee while maintaining strict confidentiality. The committee convened meetings at predetermined trial milestones, including the first week, second week, fourth week of the experiment, and one week after data collection completion, to review the accumulated data. These activities were conducted in strict compliance with applicable regulations and international standards to safeguard participant welfare and ensure the smooth progression of the trial, as well as the integrity and validity of the data. Operating independently from the sponsor and free from conflicts of interest, the committee maintained full autonomy in its oversight functions. Details of its charter may be requested from the authors 2 years after the completion of the study.

Adverse event reporting and harms {22}

The following adverse effects and corresponding management protocols were established for VR-based interventions:

1. Motion Sickness: Caused by sensory conflict between visual and vestibular systems, presenting as dizziness,nausea,or vomiting. Management: Immediate discontinuation of VR use, rest until symptom resolution, and antiemetic medication if warranted.

2. Visual Fatigue: Prolonged exposure may cause dry eyes, blurred vision, or headaches. Management: Scheduled breaks every 30 min, application of artificial tears, and ambient light optimization.

3. Musculoskeletal Discomfort: Static postures during sessions may elicit neck/back pain or muscle fatigue. Management: Postural realignment, structured stretching exercises, and physical therapy referral for persistent symptoms.

4. Psychological Distress: Hyper-realistic environments or disturbing content may trigger anxiety, fear, or mood disturbances. Management: Immediate session termination, on-site emotional support, and psychological referral for unresolved symptoms.

5. Device-Related Injuries: Environmental hazards or improper headset use may cause collisions or falls. Management: Rapid injury assessment, first-aid provision, and emergency medical evaluation for significant trauma.

Frequency and plans for auditing trial conduct {23}

A tiered auditing framework was implemented throughout the trial lifecycle: A system validation audit (covering preparatory workflows, staff responsibilities, and randomization setup) preceded first-patient enrollment. Routine monitoring visits focusing on high-risk procedures occurred after every 10 participant enrollments. Bimonthly source data verification audits scrutinized randomization systems and database integrity. Within 2 weeks after study completion, a comprehensive close-out audit examined all procedural documentation prior to final archival.

Plans for communicating important protocol amendments to relevant parties (e.g., trial participants, ethical committees) {25}

The Ethics Committee of General Hospital, Ningxia Medical University has reviewed and approved this trial protocol, confirming its ethical and scientific validity. No protocol modifications are permitted without prior approval from the Executive Committee. All amendments require formal ratification during Executive Committee meetings before sponsor notification. The Principal Investigator is responsible for timely submission of protocol deviation/violation reports. All approved changes will be promptly updated in the clinical trial.

Dissemination plans {31a}

Upon completion of the study, findings will be disseminated through publication in peer-reviewed scientific journals.

Discussion

GMI relies on action observation to activate the MNS [2], utilizing visual input to simulate movement intention and induce neuroplastic reorganization in the premotor cortex and M1 [16]. Observation (or imagination) of grasping actions enhances corticospinal excitability targeting hand muscles through activation of motor cortical networks [17]. This process essentially constitutes “offline encoding” of motor information—where the brain encodes identical movement patterns without physical execution—establishing a neuromotor template for relearning movements [18]. Post-stroke somatosensory impairments frequently disrupt motor planning [20]. The MNS responds not only to actions themselves but also to associated sensory consequences. fMRI evidence demonstrates that observing tactile stimuli activates neural networks overlapping with actual touch and motor preparation—including the primary (SI) and secondary somatosensory cortices (SII),and premotor cortex (BA6)—forming “sensorimotor shared representations” [19, 20]. During observation of touch events (e.g., object-hand/object-object interactions), SII activates selectively through automated simulation via shared neural circuitry [21]. This activation results from anticipatory encoding of goal-directed actions [22], where motor goals (e.g., stable grasping) integrate tactile properties of objects [23]. fMRI reveals enhanced functional connectivity between the dorsal premotor cortex (dPMC) and SII when observing goal-relevant sensory input [24], suggesting sensory information optimizes motor command generation. Consequently,our experimental design employs VR to deliver ecologically valid SO stimuli (e.g., visualizing icy object contact), which may drive cross-modal reorganization in somatosensory-motor integration cortices. We propose augmenting GMI therapy with SO to enhance MNS “observation-execution matching” pathways. Post-training fMRI will evaluate corresponding cortical activation patterns and connectivity changes.

Studies have demonstrated that observing videos of hand-object touch activates the primary somatosensory cortex (SI) with specificity to content [25], eliciting neural activity resembling actual tactile processing [26]. Research further indicates [27] that early enhancement of somatosensory evoked potentials (SEPs)—whether after the 30th or 60th trial—shows statistically significant positive correlations (correlation coefficient r > 0.4, corrected p < 0.05) with later behavioral improvements. Subjects exhibiting greater early increases in SI excitability subsequently demonstrated more substantial overall behavioral gains. Changes in SI excitability serve as a critical biomarker during early human motor learning,preceding alterations in motor cortex (M1) activity temporally and predicting subsequent learning efficacy. Using transcranial magnetic stimulation (TMS) to measure motor-evoked potentials, SI has been shown to modulate corticospinal output (i.e., increased MEP amplitude) via corticocortical projections influencing M1 [28]. Accordingly, our experimental design prioritizes SO before graded motor imagery training, aiming to leverage early sensory cortex activation to amplify the efficacy of subsequent motor imagery and mirror therapy. We will employ resting-state fMRI to validate the hypothesized sensory-motor cortical connectivity and determine whether modified immersive virtual reality training further enhances clinical outcomes and promotes functional brain recovery.

A limitation of this study is its inclusion of SO therapy within GMI without distinct subgroup analyses differentiating patients with versus without somatosensory impairments. Sensory function baselines were not formally assessed prior to intervention. Future research should stratify participants by pre-intervention sensory status and implement protocol-adapted SO paradigms. This approach will determine whether the combined intervention differentially promotes sensory recovery in stroke survivors with somatosensory deficits and clarify its mechanistic role in neural reorganization.

Trial status

Patient recruitment commenced in August 2025 and remains actively ongoing. Enrollment is projected to conclude by December 2025. The current trial protocol (Version 1.0, dated 21 May 2024) remains in effect throughout this recruitment phase.

Abbreviations

SO

Somatosensory observation

GMI

Graded motor imagery

VR

Virtual reality

rs-fMRI

Resting-state brain functional magnetic resonance imaging

FMA-UE

Fugl-Meyer Assessment for Upper Extremity

ARAT

Action Research Arm Test

MNS

Mirror neuron system

MT

Mirror therapy

MMSE

Mini-Mental State Examination

MBI

Modified Barthel Index

MI

Motor imagery

PreCG

Precentral gyrus

PoCG

Postcentral gyrus

SPSS

Statistical Package for the Social Sciences

CRF

Case Report Form

DMC

Data Monitoring Committee

ANOVA

Analysis of variance

LI

Laterality indices

M1

Primary motor cortex

S1

Primary somatosensory cortex

fMRI

Functional MRI

SII

Secondary somatosensory cortices

Authors’ contributions {31b}

All authors affirm accountability for the content of this research and have approved the final submitted manuscript. WS and ZN drafted the initial manuscript. ZSQ and WJK contributed to the study conception/design and performed critical manuscript revisions. MWJ, MQ, and JHM conducted systematic literature reviews and contributed to conceptual framework development. ZSQ and LRZ formulated the statistical analysis plan. WS, ZSQ, and ZN served as principal investigators, overseeing project execution and methodology implementation.

Funding {4}

This work was supported by the Key Research and Development Program of the Department of Science and Technology of Ningxia Hui Autonomous Region (Grant No.2022BEG02047).

Data availability {29}

The datasets generated and/or analyzed during this study will be made available upon reasonable request. All data-sharing proposals will undergo review by the corresponding author and require approval from the steering committee after publication of primary research findings. For inquiries regarding data access, please contact the corresponding author via email at 13995310532@163.com.

Declarations

Ethics approval and consent to participate {24}

The study protocol and informed consent documents received ethical approval from the Institutional Review Board of the General Hospital of Ningxia Medical University on 1 April 2024 (Approval No.KYLL-2024–0226). Prior to trial participation, written informed consent will be obtained from all participants along with legally authorized representatives or guardians.

Consent for publication {32}

All research participants were explicitly informed of, and consented to, the publication of study outcomes. This agreement expressly prohibits the disclosure of any individually identifiable information or personal health details in accordance with institutional privacy protocols and international data protection standards.

Competing interests {28}

The authors disclose no competing interests, whether financial or non-financial, that could influence the design, execution, or interpretation of the reported research findings.