Abstract
Background
Robot-assisted gait training is more effective in improving lower limb function and walking ability in stroke patients compared to conventional rehabilitation, but the neural mechanisms remain unclear. This study aims to explore the effects of robot-assisted gait training on lower limb motor dysfunction in stroke patients and its impact on neural activity in the motor cortex, providing objective evidence for clinical application.
Methods
Forty-two stroke patients meeting the inclusion criteria were randomly assigned to either the experimental group receiving robot-assisted gait training or the control group receiving conventional overground walking training. Assessments were conducted at baseline and after four weeks of treatment. Primary outcome measures included cortical activation measured by functional near-infrared spectroscopy (fNIRS), power ratio index (PRI), and delta/alpha power ratio (DAR) measured by quantitative electroencephalography (QEEG), and their correlation with the Fugl-Meyer Assessment (FMA) for lower limb motor function. Secondary outcome measures included FMA and Functional Ambulation Category (FAC).
Results
Data from 36 patients (18 in each group) after four weeks of treatment were analyzed. The fNIRS results indicated better activation in the premotor and supplementary motor cortices in the robot-assisted gait training group compared to the control group. QEEG analysis showed reduced PRI and DAR in the premotor, supplementary motor, and primary motor cortices in the robot-assisted gait training group, suggesting improved motor function recovery in stroke patients. Clinical scale analysis revealed superior motor function recovery in the robot-assisted gait training group compared to the control group.
Conclusions
Robot-assisted gait training significantly enhances activation in the primary motor cortex and supplementary motor area, potentially aiding stroke patients in recovering their ability to plan. PRI and DAR, particularly PRI, are valuable clinical indicators for assessing motor function recovery in stroke patients.
Trial registration
Chinese Clinical Trial Registry (ChiCTR2200060668). Registered on June 6, 2022; https://www.chictr.org.cn/showproj.html?proj=171610.
Keywords: Rehabilitation robot, Stroke, Lower limb motor dysfunction, Functional near-infrared spectroscopy, Quantitative electroencephalography
Background
Stroke is a global health issue and one of the leading causes of long-term disability. Approximately one-third of stroke patients experience permanent motor deficits, severely affecting their daily activities [1]. Lower limb motor dysfunction is a common problem among stroke patients, leading to difficulties in mobility, posture maintenance, balance, and walking. Therefore, providing rehabilitation to improve walking ability in stroke patients is necessary [2, 3].
In recent years, rehabilitation robots have become increasingly important in clinical rehabilitation [4]. Their application can relieve therapists from strenuous training tasks. By analyzing data from rehabilitation robot training, the patient’s recovery status can be assessed [5]. Due to their precision and reliability, rehabilitation robots are an effective method for improving stroke rehabilitation [6].
Currently, the neurophysiological mechanisms by which rehabilitation robots enhance functional walking ability remain unclear [7, 8]. Some scholars believe that the effectiveness of rehabilitation robots in improving functional walking ability depends on the high repetition frequency and intensity of task-oriented movements [9]. Studies have shown that conventional exercise therapy can enhance patients’ neuroplasticity [10, 11]. Compared to traditional therapy, robot-assisted gait training may more effectively promote neuroplasticity mechanisms related to motor learning and functional recovery, such as sensorimotor plasticity, effective connectivity of the frontal-parietal cortex, and interhemispheric inhibition [12].
The rise of multimodal neuroimaging technologies has significantly impacted modern neuroscience. These methods contribute independently to understanding cognitive processing [13, 14] and improving clinical diagnosis [15]. Functional near-infrared spectroscopy (fNIRS) combined with quantitative electroencephalography (QEEG) is currently favored due to its non-invasiveness, low cost, and system flexibility [16]. fNIRS is suitable for monitoring cortical activation during dynamic movement, making it possible to visualize cortical activation during dynamic movement [17]. Based on this, this research will use fNIRS to detect patients before and after robot-assisted gait training, indirectly assessing cortical neural activation by observing changes in beta values across different brain regions.
QEEG can record synchronous postsynaptic potentials of cortical neurons from the scalp [18]. The raw electroencephalography (EEG) signal is amplified, digitized, mapped, and filtered to isolate narrow frequency bands (in Hz) reflecting specific brain sources and functions, typically divided into delta (0.3–3.5 Hz), theta (4–7.5 Hz), alpha (8–13 Hz), and beta (14–30 Hz) bands. This study will use the delta/alpha ratio (DAR) and the power ratio index (PRI), which is (delta + theta)/(alpha + beta), to assess the degree of motor dysfunction and motor gain in stroke patients.
Robot-assisted gait training has been shown to effectively improve walking ability, correct abnormal gait, and promote motor function recovery and balance in hemiplegic stroke patients, but its neural mechanisms remain unclear. In this study, hemiplegic stroke patients will undergo fNIRS and QEEG assessments over a four-week period both before and after receiving robot-assisted gait training and conventional gait training, with subsequent analysis of the correlations between EEG indices and clinical outcome measures. For the first time, this research combine fNIRS and QEEG to evaluate the dynamic effects of lower limb robotic rehabilitation on the motor cortex, providing multidimensional evidence to elucidate the neuroplastic mechanisms underlying lower limb robotic therapy in stroke patients and laying a theoretical foundation for the design of personalized rehabilitation protocols.
Methods
Trial design
This trial was designed as an assessor-blinded, randomized controlled trial based on the CONSORT statement. The study was approved by the Ethics Committee of the Third Affiliated Hospital of Zhejiang Chinese Medical University (No. ZSLL-ZN-2022-011-1). The study is registered with the Chinese Clinical Trial Registry (No. ChiCTR2200060668). All subjects signed a written informed consent form before initiating the trial.
Study setting and participants.
It includes 42 patients receiving rehabilitation treatment at the Third Affiliated Hospital of Zhejiang Chinese Medical University. A computer-generated random sequence assigned patients equally to either the robot‐assisted or control group (1:1 ratio). An independent statistician, uninvolved in the study, conducted the randomization while maintaining allocation concealment until final assignment. All assessments were conducted by therapists blinded to the group assignments. The inclusion criteria were: (1) Meets stroke diagnostic criteria; (2) Aged over 18 years, first onset stroke duration of 1–6 months; (3) Lower limb modified Ashworth scale ≤ 2, Brunnstrom stage ≥ III; (4) Standing balance level II; (5) Able to walk independently for at least 15 m with assistive devices; (6) Able to understand and follow instructions; (7) No skull defects, significant brain edema, or significant brain atrophy; (8) Stable condition; (9) All patients provided informed consent signed by the patient or their legal guardian. The exclusion criteria were: (1) Severe cognitive impairment or speech disorders that prevent cooperation with training; (2) Hemianopia or visual impairments; (3) Vestibular dysfunction such as vertigo or tinnitus; (4) Severe bone and joint diseases, severe cardiopulmonary diseases preventing training; (5) Condition deteriorating, unable to continue treatment.
Sample size
This study is a randomized controlled trial using the Fugl-Meyer lower extremity motor function score as the outcome measure. According to the literature [19], it is anticipated that the intergroup difference in the Fugl-Meyer lower extremity motor function score at the end of the intervention will be 0.8 points, with a standard deviation of 0.7. A two-sided test with an α of 0.05 and a power (test efficiency) of 90% was used. Calculations determined that 16 subjects are required in each group. Considering a 10% dropout rate, a minimum of 18 subjects per group is needed, totaling at least 36 subjects for the study.
Intervention
Experimental Group: Patients undergo robot-assisted gait training using a treadmill-based rehabilitation robot (model: GR-A1). During training, patients wear a safety harness connected to the robot system for body weight support, with leg devices simulating a complete physiological gait cycle on the treadmill. Parameters such as hip and knee joint angles, leg length, and walking speed (0.5–3.0 m/s) can be adjusted. Initial training parameters are personalized and adjusted based on patient response. Training sessions last 30 min, five times a week, for a total of 20 sessions.
Fig. 1.
Robotic-assisted gait training setup
Control Group: Patients receive conventional overground walking training, walking back and forth in a flat indoor corridor until reaching a Borg perceived exertion level of 4. After resting, patients resume training following the same procedure until the session ends. Each session lasts 30 min, five times a week, for a total of 20 sessions.
fNIRS data processing and analysis
All selected patients were evaluated in an awake and quiet state by personnel who were professionally and uniformly trained according to standard operating procedures. Evaluations were conducted one day before treatment and four weeks after treatment for both the experimental and control groups to ensure consistency in assessment time points, allowing for accurate comparison of pre- and post-treatment effects.
fNIRS: Using a portable functional near-infrared spectroscopy (fNIRS) brain imaging device from Japan (LIGHTNIRS), which has 22 channels, including 8 transmitters and 8 receivers, the detection is performed. The data collection environment should be quiet, with subdued lighting, and free from electromagnetic interference. The electrodes are attached to the corresponding scalp positions, and the patient walks in a straight line for 20 s, then rests for 20 s, repeating this measurement five times while maintaining an upright posture during walking. The primary areas of data collection include the primary motor cortex, premotor cortex, and supplementary motor cortex, observing hemodynamic changes in the cortex until stable signals of oxygenated hemoglobin concentration changes are obtained. Thus, the change in oxygenated hemoglobin for each channel is determined. The coordinate information for each channel is confirmed using a 3D positioning system on a standard head model to activate the cortical areas of the corresponding channels. According to the modified Beer-Lambert law, the change in hemoglobin level can be quantified by the β value of fNIRS, which indirectly reflects neural activation in various cortical areas.
The target localization was performed using the FASTRAK digital three-dimensional digitizer (Polhemus USA) to ensure sufficient precision for measuring the activity of the primary motor cortex, premotor cortex, and supplementary motor area (Fig. 2).
Fig. 2.
fNIRS channel positions overlaid on a 3D brain model
This study used the NIRS_KIT software based on Matlab 2021a (MathWorks Inc., Natick, MA, USA) to remove artifacts and preprocess the raw intensity data of fNIRS [20]. To reduce errors, this research excluded the data from the first and last task periods and selected data from the middle three task periods for analysis to obtain beta values reflecting the corresponding cortical activation levels. This research used the beta values before treatment as baseline (beta1), subtracted beta1 from the beta values after treatment (beta2), and obtained the beta change values (Δbeta = beta2 - beta1) after treatment. Subsequently, this research used statistical methods such as two-sample t-tests, rank sum tests, etc., to analyze the cortical activation of each channel in the experimental and control groups. This research generated heat maps and placed them on 3D brain models to visually demonstrate changes in cortical activation before and after treatment [21].
QEEG data processing and analysis
QEEG: One day before and four weeks after treatment, researcher conducted QEEG examinations on the patients. The QEEG data collection environment was kept quiet and free from electromagnetic interference. The German QEEG device (actiCHamp) was used, and after confirming the connection between the EEG software and the relevant acquisition devices, electrodes were placed on the corresponding scalp positions according to the international 10–20 system. Medical conductive gel was applied to achieve appropriate impedance levels. Patients were comfortably seated with eyes closed to maintain an awake state and avoid movement during the assessment. The main areas of data collection included the primary motor cortex (Cz, C3, C4), premotor cortex (FC1, FC2), and supplementary motor area (FCz), with each data collection session lasting three minutes to obtain stable signals of EEG parameters (absolute power in δ, θ, α, and β frequency bands) [22].
After data collection, the QEEG data were processed and analyzed for artifact removal and preprocessing using the original QEEG analyzer software. To minimize errors, EEG data from the first minute and the last minute were excluded, and data from the middle minute were selected for analysis to extract the absolute power values of each frequency band. Subsequently, the DAR (Delta Absolute Ratio) and PRI (Peak Ratio Index) were calculated for each patient, with the data before treatment as baseline (DAR1 and PRI1). The values after treatment (DAR2 and PRI2) were subtracted from the baseline values to obtain the changes in DAR and PRI after treatment (ΔDAR = DAR2 - DAR1, ΔPRI = PRI2 - PRI1). Finally, statistical methods such as two-sample t-tests, non-parametric tests, etc., were used to analyze the data. Using R (version 3.5.3), correlations between the ΔDAR and ΔPRI values of the premotor cortex, supplementary motor area, and primary motor cortex of each patient and the changes in the Fugl-Meyer Lower Extremity Assessment (FMA) scores were explored to investigate the relationship between treatment effects and neurophysiological indicators.
Clinical assessment scales
Functional Ambulation Classification (FAC): A scale consisting of six grades designed to classify the level of physical support required for safe walking by subjects, ranging from Level 0 (unable to walk without the help of two people) to Level 5 (independent walking on uneven surfaces and stairs) [23].
Fugl-Meyer Lower Extremity Assessment (FMA): Consists of 17 items, each scored out of two points. Evaluates the patient’s lower limb motor function, including exaggerated reflexes, synergistic movement of extensor and flexor muscles, movements without synergistic coordination, coordination speed, and ability, among others [24].
Outcome measures
At the beginning stage of the study, researcher collected medical and demographic information as shown in Table 1. The study included fNIRS and QEEG examinations of participants before treatment and after four weeks of rehabilitation therapy, along with clinical scale assessments.
Table 1.
Baseline characteristics of study participants
| Characteristic | Experimental Group | Control Group | P |
|---|---|---|---|
| Age(years) | 57.83 ± 13.85 | 63.22 ± 13.53 | 0.246 |
| Male/ female | 13/5 | 14/4 | 0.700 |
| Months post stroke | 2.50 ± 1.79 | 2.11 ± 1.08 | 0.791 |
|
Disease site (left/ right) |
11/7 | 10/8 | 0.735 |
|
Stroke type (hemorrhage/infarction) |
4/14 | 4/14 | 1.000 |
| FMA | 21.61 ± 4.41 | 20.06 ± 3.17 | 0.304 |
| FAC | 2.78 ± 0.73 | 2.78 ± 0.65 | 0.944 |
Values denote means ± SD unless specified otherwise
Primary outcomes: Changes in cerebral cortical oxygenated hemoglobin measured by fNIRS (Δbeta); changes in DAR and PRI measured by QEEG after treatment; correlation between changes in DAR and PRI after treatment and the Fugl-Meyer Lower Extremity Assessment (FMA). Secondary outcomes: Functional Ambulation Classification (FAC); FMA scores.
There were no statistically significant differences between the two groups of patients in terms of gender, age, duration of illness, affected side (left/right), nature of onset (hemorrhage/infarction), and pre-treatment FMA and FAC scores (P > 0.05).
Statistical analysis
Statistical analysis was performed using SPSS 25.0 software. All data passed normality tests. Continuous variables are presented as mean ± standard deviation (X ± s). Group comparisons were conducted using two-sample t-tests and non-parametric rank-sum tests, while paired t-tests were used for within-group comparisons before and after treatment. A significance level of P < 0.05 was considered statistically significant. For correlation analysis, “R” (version 3.5.3) was used to conduct linear correlation analysis, with correlation coefficients represented by r and P < 0.05 considered statistically significant.
Results
The study included 42 patients from June 2022 to March 2023. During the study, five patients withdrew from assessment due to transfer to another hospital or discharge home, and one patient withdrew due to unwillingness to continue treatment. In the experimental group, three patients deviated from the protocol: two were transferred to another hospital or discharge home and one declined to continue treatment. In the control group, all three patients were transferred to another facility. Ultimately, 36 patients completed the entire study, with 18 patients in the experimental group and 18 patients in the control group (Fig. 3).
Fig. 3.
Flow diagram of the study
In the experimental group, the Δbeta values of channels 1, 8, 9, 11, 12, 18, 19, 20, and 21 showed a significant increase compared to the control group with statistical significance (P < 0.05) (Figs. 4 and 5). According to the three-dimensional digitizer, channels with statistically significant differences were mainly distributed in the premotor cortex and supplementary motor area (channels 1, 8, 9, 11, 12, 18, 19, 20) with only one channel located in the primary motor cortex (channel 21) (Table 2).
Fig. 4.
Cortical activation patterns in the experimental group
Fig. 5.
Cortical activation patterns in the control group
Table 2.
Changes in cortical activation (Δbeta) measured by fNIRS across different channels
| Channel | Experimental Group | Control Group | P |
|---|---|---|---|
| 1 | 0.0027 ± 0.0044 | -0.0006 ± 0.0031 | 0.012* |
| 2 | 0.0028 ± 0.0048 | 0.0017 ± 0.0030 | 0.448 |
| 3 | 0.0026 ± 0.0075 | 0.0022 ± 0.0055 | 0.874 |
| 4 | 0.0028 ± 0.0081 | 0.0002 ± 0.0062 | 0.184 |
| 5 | 0.0018 ± 0.0074 | -0.0001 ± 0.0049 | 0.506 |
| 6 | 0.0024 ± 0.0040 | 0.0011 ± 0.0022 | 0.216 |
| 7 | 0.0016 ± 0.0036 | 0.0025 ± 0.0048 | 0.899 |
| 8 | 0.0024 ± 0.0032 | 0.0002 ± 0.0027 | 0.031* |
| 9 | 0.0049 ± 0.0060 | -0.0008 ± 0.0027 | 0.002* |
| 10 | 0.0012 ± 0.0063 | 0.0019 ± 0.0032 | 0.975 |
| 11 | 0.0087 ± 0.0132 | -0.0046 ± 0.0128 | 0.008* |
| 12 | 0.0048 ± 0.0070 | -0.0024 ± 0.0066 | 0.002* |
| 13 | 0.0040 ± 0.0044 | 0.0012 ± 0.0038 | 0.114 |
| 14 | 0.0022 ± 0.0036 | 0.0011 ± 0.0028 | 0.305 |
| 15 | 0.0056 ± 0.0082 | 0.0023 ± 0.0086 | 0.174 |
| 16 | 0.0021 ± 0.0035 | 0.0008 ± 0.0032 | 0.243 |
| 17 | 0.0037 ± 0.0049 | 0.0014 ± 0.0027 | 0.174 |
| 18 | 0.0024 ± 0.0043 | -0.0005 ± 0.0034 | 0.034* |
| 19 | 0.0056 ± 0.0110 | -0.0009 ± 0.0084 | 0.040* |
| 20 | 0.0029 ± 0.0033 | -0.00003 ± 0.0031 | 0.029* |
| 21 | 0.0039 ± 0.0071 | 0.0015 ± 0.0030 | 0.023* |
| 22 | 0.0018 ± 0.0032 | 0.0004 ± 0.0039 | 0.253 |
*Indicates statistical significance (P < 0.05)
Due to the time-intensive nature of QEEG assessments and limited equipment availability, this research randomly selected nine patients from both the experimental and control groups, resulting in a total of 18 participants, who underwent QEEG evaluations before and after the intervention. The analysis results showed that ΔPRI in the primary motor cortex (M1), premotor cortex (PMC), and supplementary motor area (SMA) were significantly reduced in the experimental group compared to the control group with statistical significance (Table 3). ΔDAR in the primary motor cortex and premotor cortex was significantly reduced in the experimental group compared to the control group with statistical significance (Table 4).
Table 3.
Inter-group comparison of ΔPRI in various regions of the motor cortex before and after treatment
| Motor cortex | Experimental Group | Control Group | P |
|---|---|---|---|
| M1 | -4.361 ± 8.266 | -4.199 ± 0.798 | 0.038* |
| PMC | -10.426 ± 17.574 | -0.526 ± 0.723 | 0.007* |
| SMA | -9.703 ± 13.590 | -0.700 ± 1.172 | 0.015* |
Table 4.
Inter-group comparison of ΔDAR in various regions of the motor cortex before and after treatment
| Motor cortex | Experimental Group | Control Group | P |
|---|---|---|---|
| M1 | -6.482 ± 12.542 | -0.511 ± 0.721 | 0.038* |
| PMC | -15.781 ± 27.464 | -0.628 ± 0.666 | 0.047* |
| SMA | -14.415 ± 21.656 | 0.877 ± 1.834 | 0.102 |
This research conducted a correlation analysis between ΔPRI in the primary motor cortex, premotor cortex, supplementary motor area, and changes in Fugl-Meyer Assessment (ΔFMA) scores. The results showed that ΔPRI before and after treatment in these areas were correlated with ΔFMA scores with statistical significance (Fig. 6).
Fig. 6.
Correlation between ΔPRI and changes in FMA scores. (a) Correlation between ΔPRI and ΔFMA of M1; (b) Correlation between ΔPRI and ΔFMA of PMC; (c) Correlation between ΔPRI and ΔFMA of SMA
This research conducted a correlation analysis between ΔDAR in the primary motor cortex, premotor cortex, supplementary motor area, and changes in Fugl-Meyer Assessment (ΔFMA). This research found that ΔDAR before and after treatment in the primary motor cortex and supplementary motor area were correlated with ΔFMA scores with statistical significance (Fig. 7).
Fig. 7.
Correlation between ΔDAR and changes in FMA scores. (a) Correlation between ΔDAR and ΔFMA of M1; (b) Correlation between ΔDAR and ΔFMA of PMC; (c) Correlation between ΔDAR and ΔFMA of SMA
Through clinical scale assessments, it was found that after treatment, patients in the experimental group showed significant improvement in Fugl-Meyer Assessment (FMA) scores and Functional Ambulation Classification (FAC) scores. The FAC scores of patients in the control group also showed significant improvement. Compared to the control group, the improvement in FMA and FAC scores was more significant in the experimental group, and these differences were statistically significant (Table 5).
Table 5.
Comparison of FMA scores and FAC scores before and after treatment in both groups
| Experimental Group | Control Group | P -value | |||||
|---|---|---|---|---|---|---|---|
| Intragroup difference | Interaction effect |
||||||
| Week0 | Week4 | Week0 | Week4 | Experimental Group | Control Group |
||
| FMA | 21.61 ± 4.41 | 26.06 ± 3.72 | 20.06 ± 3.17 | 21.33 ± 3.13 | 0.001* | 0.105 | <0.001* |
| FAC | 2.78 ± 0.73 | 4.11 ± 0.68 | 2.78 ± 0.65 | 3.28 ± 0.75 | <0.001* | 0.040* | 0.002* |
Discussion
In recent years, rehabilitation robotics have shown tremendous potential in promoting lower limb motor function recovery in stroke patients, although their neural mechanisms remain unclear. This study employed a combined approach of neurophysiological and clinical assessments to delve deeper into the therapeutic effects of rehabilitation robotics and their underlying neural mechanisms.
This research utilized the FMA and FAC scales to evaluate the effects of robotic-assisted gait training on improving walking function in stroke patients. This research applied fNIRS and QEEG analyses to investigate the effects of robotic-assisted gait training on motor cortical neural activity in stroke patients. Compared to the control group receiving conventional gait training, patients undergoing robotic-assisted gait training showed significantly greater improvements in lower limb motor function and walking ability. fNIRS indicated significant activation in the premotor cortex and supplementary motor area in the experimental group compared to the control group. QEEG revealed significantly decreased Delta Absolute Ratio (DAR) and Peak Ratio Index (PRI) in the motor cortex of the experimental group compared to the control group. Moreover, PRI showed a higher correlation with FMA scores than DAR, suggesting that PRI may be more suitable for predicting motor function recovery outcomes.
The FMA scale is widely used to evaluate motor impairments following stroke and is globally recognized as a critical assessment tool in clinical practice and research [25, 26]. This research used the FMA and FAC scales to evaluate patients undergoing rehabilitation robotic-assisted gait training compared to those receiving conventional gait training. The results showed that patients undergoing rehabilitation robotic-assisted gait training had significantly better recovery outcomes compared to the control group, suggesting that robotic-assisted gait training is crucial in promoting lower limb motor function recovery in stroke patients. These findings are consistent with previous literature [27, 28]. Prior studies [29] have demonstrated that robot-assisted training more effectively activates the motor cortex and promotes neural remodeling, thereby enhancing motor coordination and gait stability. The clinical data from our study further substantiate this observation.
Stroke-induced motor impairments are primarily associated with the pathophysiology affecting different regions of the motor cortex [30, 31]. The motor cortex is a critical area in the cerebral cortex involved in the planning, control, and execution of voluntary movements [32]. It consists primarily of the primary motor cortex, the premotor cortex, and the supplementary motor area [33]. The primary motor cortex is the main region responsible for generating neural impulses that are transmitted to the spinal cord to control body movements [34]. The premotor cortex is involved in movement preparation, sensory integration, and spatial navigation, particularly in the direct control of proximal and trunk muscles [35]. The supplementary motor area is closely associated with motor planning and is primarily involved in the generation and control of voluntary movements [36]. Analysis using fNIRS has shown significant enhancement in neural activation in the premotor cortex and supplementary motor area during walking after rehabilitation robotic-assisted gait training. This indicates that rehabilitation robotic training is crucial for restoring motor planning and control abilities in stroke patients.
In recent years, QEEG has been widely used to assess brain function and identify biomarkers associated with brain injury and recovery. In stroke patients, QEEG power is significantly affected, characterized by increased δ band power and decreased α and β band powers, resulting in a diffuse slow-wave QEEG pattern [37]. The increase in slow waves and decrease in fast waves are directly related to neuronal metabolism and reflect ischemic injury [38]. Analysis using QEEG can provide valuable information about the recovery of motor function in stroke patients, aiding in the personalized development of rehabilitation plans [39].
Our study utilized QEEG as a non-invasive assessment method with high temporal resolution, facilitating rapid evaluation of brain function [40]. Power changes in specific frequency bands are closely related to brain functional status, allowing QEEG to sensitively detect typical QEEG abnormalities in stroke patients [41]. DAR and PRI, commonly used QEEG parameters, are considered potential indicators for predicting the severity of post-stroke functional impairments. In studies by VAN et al. [42], PRI was found to be closely associated with upper limb motor recovery post-stroke, suggesting that PRI may be more sensitive in assessing rehabilitation outcomes. Another study [43] found that DAR and PRI were negatively correlated with motor recovery in stroke patients before and after rehabilitation. Our research observed that stroke patients undergoing rehabilitation robot-assisted walking training showed significant reductions in PRI in the primary motor cortex, premotor cortex, and supplementary motor area compared to the control group. Furthermore, the changes in these PRI values were significantly negatively correlated with ΔFMA. In contrast, ΔDAR showed significant negative correlations only with ΔFMA in the supplementary motor cortex and primary motor cortex. These findings suggest that PRI may be a more sensitive indicator for assessing gains in post-stroke rehabilitation of walking function, consistent with the findings of Trujillo [44].
Using combined fNIRS and QEEG monitoring, this study revealed that robot-assisted gait training significantly enhances neural activation in the premotor cortex and supplementary motor area, while electrophysiological indices such as PRI and DAR show a marked reduction. These findings suggest that robotic training facilitates neuroplasticity, thereby accelerating motor function recovery. Compared with conventional rehabilitation approaches, this mechanism highlights the unique role of robotic-assisted training in modulating neural activity, providing new experimental evidence for post-stroke neural remodeling. Moreover, our results offer robust theoretical support for clinical translation. Objective monitoring of electrophysiological changes not only enables a more precise assessment of rehabilitation progress but also provides quantifiable metrics for designing personalized rehabilitation strategies. In the future, an evaluation system integrating fNIRS and QEEG data may allow real-time tracking of neural recovery in clinical settings, guiding the dynamic adjustment of rehabilitation protocols. This approach holds promise for achieving precision, personalization, and enhanced therapeutic outcomes in stroke rehabilitation.
Limitations
This study has two main limitations: (1) It is a small-sample study without follow-up, so conclusions drawn need caution. Further research could consider larger sample sizes and multi-center clinical trials with longer treatment and follow-up periods. (2) The study did not classify subjects based on the type and location of stroke, leading to potential heterogeneity among patients. Future studies could involve stratified classification to provide more personalized research outcomes and treatment strategies.
Conclusion
Robotic-assisted gait training effectively promotes lower limb motor recovery in stroke patients compared to conventional overground walking training. fNIRS analysis revealed increased activation in the premotor cortex and supplementary motor area, suggesting enhanced motor planning and execution. QEEG findings demonstrated significant reductions in PRI and DAR, which were negatively correlated with FMA scores, indicating their potential as biomarkers for motor function recovery. These results provide neurophysiological evidence supporting the role of robotic-assisted rehabilitation in facilitating neuroplasticity. Future studies should explore long-term outcomes and optimize rehabilitation protocols to enhance clinical efficacy.
Author contributions
XL, HHZ and JEC were involved in the development and design of the study concept; WYZ, JNW and LD were responsible for intervention and assessment; XL, WYZ and LD were in charge of data acquisition and analysis; XL, NSL, TFJ and LG contributed to the initial manuscript writing. All authors revised and agreed to the final version of this article. XL, HHZ and WYZ contributed equally to this work.
Funding
Key Scientific Research and Development Program of Zhejiang Province, PRC (No.2021C03050); General Scientific Research Project of Zhejiang Department of Education, PRC (Y202145964); Hangzhou Science and Technology Development Plan Project, PRC (2020ZDSJ0881).
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethical approval
The study was approved by the Ethics Committee of the Third Affiliated Hospital of Zhejiang Chinese Medical University (No. ZSLL-ZN-2022-011-1). The study is registered with the Chinese Clinical Trial Registry (No. ChiCTR2200060668). All subjects signed a written informed consent form before initiating the trial.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Harvey RL. Predictors of functional outcome following stroke. Phys Med Rehabil Clin North Am. 2015;26(4):583–98. 10.1016/j.pmr.2015.07.002. [DOI] [PubMed]
- 2.Törnbom K, Danielsson A. Experiences of treadmill walking with non-immersive virtual reality after stroke or acquired brain injury - A qualitative study. PLoS ONE. 2018;13 12:e0209214. 10.1371/journal.pone.0209214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Beyaert C, Vasa R, Frykberg GE. Gait post-stroke: pathophysiology and rehabilitation strategies. Neurophysiologie clinique = Clin Neurophysiol. 2015;45:4–5:335 − 55 doi: 10.1016/j.neucli.2015.09.005. [DOI] [PubMed]
- 4.Zhang B, Wong KP, Kang R, Fu S, Qin J, Xiao Q. Efficacy of Robot-Assisted and virtual reality interventions on balance, gait, and daily function in patients with stroke: A systematic review and network Meta-analysis. Arch Phys Med Rehabil. 2023;104 10:1711–9. 10.1016/j.apmr.2023.04.005. [DOI] [PubMed] [Google Scholar]
- 5.Louie DR, Mortenson WB, Durocher M, Teasell R, Yao J, Eng JJ. Exoskeleton for post-stroke recovery of ambulation (ExStRA): study protocol for a mixed-methods study investigating the efficacy and acceptance of an exoskeleton-based physical therapy program during stroke inpatient rehabilitation. BMC Neurol. 2020;20(1:35). 10.1186/s12883-020-1617-7. [DOI] [PMC free article] [PubMed]
- 6.Zuccon G, Lenzo B, Bottin M, Rosati G. Rehabilitation robotics after stroke: a bibliometric literature review. Expert Rev Med Dev. 2022;19 5:405–21. 10.1080/17434440.2022.2096438. [DOI] [PubMed] [Google Scholar]
- 7.Yang HE, Kyeong S, Lee SH, Lee WJ, Ha SW, Kim SM, et al. Structural and functional improvements due to robot-assisted gait training in the stroke-injured brain. Neurosci Lett. 2017;637:114–9. 10.1016/j.neulet.2016.11.039. [DOI] [PubMed] [Google Scholar]
- 8.Kim DH, Kang CS, Kyeong S. Robot-assisted gait training promotes brain reorganization after stroke: A randomized controlled pilot study. NeuroRehabilitation. 2020;46 4:483–9. 10.3233/nre-203054. [DOI] [PubMed] [Google Scholar]
- 9.Khalid YM, Gouwanda D, Parasuraman S. A review on the mechanical design elements of ankle rehabilitation robot. Proceedings of the Institution of Mechanical Engineers Part H, Journal of engineering in medicine. 2015;2296:452– 63; 10.1177/0954411915585597 [DOI] [PubMed]
- 10.Kwakkel G, Kollen BJ, Krebs HI. Effects of robot-assisted therapy on upper limb recovery after stroke: a systematic review. Neurorehabilit Neural Repair. 2008;22(2):111–21. 10.1177/1545968307305457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Klamroth-Marganska V. Stroke rehabilitation: therapy robots and assistive devices. Adv Exp Med Biol. 2018;1065:579–87. 10.1007/978-3-319-77932-4_35. [DOI] [PubMed] [Google Scholar]
- 12.Cho JE, Lee WH, Shin JH, Kim H. Effects of bi-axial ankle strengthening on muscle co-contraction during gait in chronic stroke patients: A randomized controlled pilot study. Gait Posture. 2021;87:177–83. 10.1016/j.gaitpost.2021.04.011. [DOI] [PubMed] [Google Scholar]
- 13.Calabrò RS, Cacciola A, Bertè F, Manuli A, Leo A, Bramanti A et al. Robotic gait rehabilitation and substitution devices in neurological disorders: where are we now? neurological sciences: official journal of the Italian neurological society and of the Italian society of clinical neurophysiology. 2016;37 4:503–14; 10.1007/s10072-016-2474-4 [DOI] [PubMed]
- 14.Uçar DE, Paker N, Buğdaycı D. Lokomat: a therapeutic chance for patients with chronic hemiplegia. NeuroRehabilitation. 2014;34 3:447–53. 10.3233/nre-141054. [DOI] [PubMed] [Google Scholar]
- 15.Beer S, Aschbacher B, Manoglou D, Gamper E, Kool J, Kesselring J. Robot-assisted gait training in multiple sclerosis: a pilot randomized trial. Mult Scler. 2008;14(2):231–6. 10.1177/1352458507082358. [DOI] [PubMed] [Google Scholar]
- 16.Cho KH, Park SJ. Effects of joint mobilization and stretching on the range of motion for ankle joint and Spatiotemporal gait variables in stroke patients. J Stroke Cerebrovasc Diseases: Official J Natl Stroke Association. 2020;29 8:104933. 10.1016/j.jstrokecerebrovasdis.2020.104933. [DOI] [PubMed] [Google Scholar]
- 17.Wagner JM, Kremer TR, Van Dillen LR, Naismith RT. Plantarflexor weakness negatively impacts walking in persons with multiple sclerosis more than plantarflexor spasticity. Arch Phys Med Rehabil. 2014;95 7:1358–65. 10.1016/j.apmr.2014.01.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Brito R, Baltar A, Berenguer-Rocha M, Shirahige L, Rocha S, Fonseca A, et al. Intrahemispheric EEG: A new perspective for quantitative EEG assessment in poststroke individuals. Neural Plast. 2021;2021:5664647. 10.1155/2021/5664647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mustafaoglu R, Erhan B, Yeldan I, Gunduz B, Tarakci E. Does robot-assisted gait training improve mobility, activities of daily living and quality of life in stroke? A single-blinded, randomized controlled trial. Acta Neurol Belgica. 2020;120(2):335–44. 10.1007/s13760-020-01276-8. [DOI] [PubMed] [Google Scholar]
- 20.Fishburn FA, Ludlum RS, Vaidya CJ, Medvedev AV. Temporal derivative distribution repair (TDDR): A motion correction method for fNIRS. NeuroImage. 2019;184:171–9. 10.1016/j.neuroimage.2018.09.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sun PP, Tan FL, Zhang Z, Jiang YH, Zhao Y, Zhu CZ. Feasibility of functional Near-Infrared spectroscopy (fNIRS) to investigate the mirror neuron system: an experimental study in a Real-Life situation. Front Hum Neurosci. 2018;12:86. 10.3389/fnhum.2018.00086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Xiao R, Shida-Tokeshi J, Vanderbilt DL, Smith BA. Electroencephalography power and coherence changes with age and motor skill development across the first half year of life. PLoS ONE. 2018;13(1):e0190276. 10.1371/journal.pone.0190276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Moucheboeuf G, Griffier R, Gasq D, Glize B, Bouyer L, Dehail P, et al. Effects of robotic gait training after stroke: A meta-analysis. Annals Phys Rehabilitation Med. 2020;63 6:518–34. 10.1016/j.rehab.2020.02.008. [DOI] [PubMed] [Google Scholar]
- 24.Dorsch S, Carling C, Cao Z, Fanayan E, Graham PL, McCluskey A, et al. Bobath therapy is inferior to task-specific training and not superior to other interventions in improving arm activity and arm strength outcomes after stroke: a systematic review. J Physiotherapy. 2023;69(1):15–22. 10.1016/j.jphys.2022.11.008. [DOI] [PubMed] [Google Scholar]
- 25.Kwakkel G, Lannin NA, Borschmann K, English C, Ali M, Churilov L, et al. Standardized measurement of sensorimotor recovery in stroke trials: Consensus-based core recommendations from the stroke recovery and rehabilitation roundtable. Int J Stroke. 2017;12 5:451–61. 10.1177/1747493017711813. [DOI] [PubMed] [Google Scholar]
- 26.Bushnell C, Bettger JP, Cockroft KM, Cramer SC, Edelen MO, Hanley D, et al. Chronic stroke outcome measures for motor function intervention trials: expert panel recommendations. Circ Cardiovasc Qual Outcomes. 2015;8(6 Suppl 3):S163–9. 10.1161/circoutcomes.115.002098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lin YN, Huang SW, Kuan YC, Chen HC, Jian WS, Lin LF. Hybrid robot-assisted gait training for motor function in subacute stroke: a single-blind randomized controlled trial. J Neuroeng Rehabil. 2022;19(1:99). 10.1186/s12984-022-01076-6. [DOI] [PMC free article] [PubMed]
- 28.Yoo HJ, Bae CR, Jeong H, Ko MH, Kang YK, Pyun SB. Clinical efficacy of overground powered exoskeleton for gait training in patients with subacute stroke: A randomized controlled pilot trial. Medicine. 2023;102(4):e32761. 10.1097/md.0000000000032761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kim H, Park G, Shin JH, You JH. Neuroplastic effects of end-effector robotic gait training for hemiparetic stroke: a randomised controlled trial. Sci Rep. 2020;10(1:12461). 10.1038/s41598-020-69367-3. [DOI] [PMC free article] [PubMed]
- 30.Hordacre B, Moezzi B, Ridding MC. Neuroplasticity and network connectivity of the motor cortex following stroke: A transcranial direct current stimulation study. Hum Brain Mapp. 2018;39 8:3326–39. 10.1002/hbm.24079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sanders ZB, Fleming MK, Smejka T, Marzolla MC, Zich C, Rieger SW, et al. Self-modulation of motor cortex activity after stroke: a randomized controlled trial. Brain. 2022;145 10:3391–404. 10.1093/brain/awac239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Svoboda K, Li N. Neural mechanisms of movement planning: motor cortex and beyond. Curr Opin Neurobiol. 2018;49:33–41. 10.1016/j.conb.2017.10.023. [DOI] [PubMed] [Google Scholar]
- 33.Park J. Movement disorders following cerebrovascular lesion in the basal ganglia circuit. J Mov Disord. 2016;9 2:71–9. 10.14802/jmd.16005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Cheng S, Xin R, Zhao Y, Wang P, Feng W, Liu P. Evaluation of fMRI activation in post-stroke patients with movement disorders after repetitive transcranial magnetic stimulation: a scoping review. Front Neurol. 2023;14:1192545. 10.3389/fneur.2023.1192545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wang Q, Dai W, Xu S, Zhu S, Sui Y, Kan C, et al. Brain activation of the PFC during dual-task walking in stroke patients: A systematic review and meta-analysis of functional near-infrared spectroscopy studies. Front NeuroSci. 2023;17:1111274. 10.3389/fnins.2023.1111274. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Wu J, Cheng H, Zhang J, Bai Z, Cai S. The modulatory effects of bilateral arm training (BAT) on the brain in stroke patients: a systematic review. Neurol Sciences: Official J Italian Neurol Soc Italian Soc Clin Neurophysiol. 2021;42 2:501–11. 10.1007/s10072-020-04854-z. [DOI] [PubMed] [Google Scholar]
- 37.Fanciullacci C, Panarese A, Spina V, Lassi M, Mazzoni A, Artoni F, et al. Connectivity measures differentiate cortical and subcortical Sub-Acute ischemic stroke patients. Front Hum Neurosci. 2021;15:669915. 10.3389/fnhum.2021.669915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Wu J, Srinivasan R, Burke Quinlan E, Solodkin A, Small SL, Cramer SC. Utility of EEG measures of brain function in patients with acute stroke. J Neurophysiol. 2016;115 5:2399–405. 10.1152/jn.00978.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Mahmoud RMA, Basheer NF, Elzanaty MAHMZ, Ragab MY. Impact of somatosensory training on neural and functional recovery of lower extremity in patients with chronic stroke: A single blind controlled randomized trial. Int J Environ Res Public Health. 2021;18(2). 10.3390/ijerph18020583. [DOI] [PMC free article] [PubMed]
- 40.Hallett M, DelRosso LM, Elble R, Ferri R, Horak FB, Lehericy S, et al. Evaluation of movement and brain activity. Clin Neurophysiol. 2021;132 10:2608–38. 10.1016/j.clinph.2021.04.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Bentes C, Peralta AR, Viana P, Martins H, Morgado C, Casimiro C, et al. Quantitative EEG and functional outcome following acute ischemic stroke. Clin Neurophysiology: Official J Int Federation Clin Neurophysiol. 2018;129 8:1680–7. 10.1016/j.clinph.2018.05.021. [DOI] [PubMed] [Google Scholar]
- 42.Van Criekinge T, Heremans C, Burridge J, Deutsch JE, Hammerbeck U, Hollands K, et al. Standardized measurement of balance and mobility post-stroke: Consensus-based core recommendations from the third stroke recovery and rehabilitation roundtable. Neurorehabilit Neural Repair. 2024;38 1:41–51. 10.1177/15459683231209154. [DOI] [PubMed] [Google Scholar]
- 43.Mane R, Chew E, Phua KS, Ang KK, Robinson N, Vinod AP, et al. Prognostic and monitory EEG-Biomarkers for BCI Upper-Limb stroke rehabilitation. IEEE Trans Neural Syst Rehabilitation Engineering: Publication IEEE Eng Med Biology Soc. 2019;27 8:1654–64. 10.1109/tnsre.2019.2924742. [DOI] [PubMed] [Google Scholar]
- 44.Trujillo P, Mastropietro A, Scano A, Chiavenna A, Mrakic-Sposta S, Caimmi M, et al. Quantitative EEG for predicting upper limb motor recovery in chronic stroke Robot-Assisted rehabilitation. IEEE Trans Neural Syst Rehabilitation Engineering: Publication IEEE Eng Med Biology Soc. 2017;25 7:1058–67. 10.1109/tnsre.2017.2678161. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.