Objective: This single-arm multisite trial investigates the efficacy of the Neurostyle Brain Exercise Therapy Towards Enhanced Recovery (nBETTER) system, an Electroencephalogram (EEG)-based Motor Imagery Brain-Computer Interface (MI-BCI) employing visual feedback, for upper-limb stroke rehabilitation, and the presence of EEG correlates of mental fatigue during BCI usage. Methods: Thirteen recruited stroke patients underwent thrice-weekly nBETTER therapy coupled with standard arm therapy over 6 weeks. Upper-extremity Fugl-Meyer Motor Assessment (FMA) scores were measured at baseline (Week 0), post-intervention (Week 6) and follow-ups (Weeks 12 and 24). In total, 11/13 patients (mean age 55.2 years old, mean post-stroke duration 333.7 days, mean baseline FMA 35.5) completed the study. Results: Significant FMA gains relative to baseline were observed at Weeks 6 and 24. Retrospectively comparing to the standard arm therapy (SAT) control group and BCI with haptic knob (BCI-HK) intervention group from a previous similar study, the SAT group had no significant gains whereas the BCI-HK group had significant gains at Weeks 6, 12 and 24. EEG analysis revealed significant positive correlations between relative beta power and BCI performance in the frontal and central brain regions, suggesting that mental fatigue may contribute to poorer BCI performance. Conclusion: nBETTER, an EEG-based MI-BCI employing only visual feedback, helps stroke survivors sustain short-term FMA improvement. Analysis of EEG relative beta power indicates that mental fatigue may be present. Significance: This study adds nBETTER to the growing literature of safe and effective stroke rehabilitation MI-BCI, and suggests an additional fatigue-monitoring role in future such BCI.

Rehabilitation of hand function is challenging, and only few studies have investigated robot-assisted rehabilitation focusing on distal joints of the upper limb. This paper investigates the feasibility of using the HapticKnob, a table-top end-effector device, for robot-assisted rehabilitation of grasping and forearm pronation/supination, two important functions for activities of daily living involving the hand, and which are often impaired in chronic stroke patients. It evaluates the effectiveness of this device for improving hand function and the transfer of improvement to arm function. A single group of fifteen chronic stroke patients with impaired arm and hand functions (Fugl-Meyer motor assessment scale (FM) 10-45/66) participated in a 6-week 3-hours/week rehabilitation program with the HapticKnob. Outcome measures consisted primarily of the FM and Motricity Index (MI) and their respective subsections related to distal and proximal arm function, and were assessed at the beginning, end of treatment and in a 6-weeks follow-up. Thirteen subjects successfully completed robot-assisted therapy, with significantly improved hand and arm motor functions, demonstrated by an average 3.00 points increase on the FM and 4.55 on the MI at the completion of the therapy (4.85 FM and 6.84 MI six weeks post-therapy). Improvements were observed both in distal and proximal components of the clinical scales at the completion of the study (2.00 FM wrist/hand, 2.55 FM shoulder/elbow, 2.23 MI hand and 4.23 MI shoulder/elbow). In addition, improvements in hand function were observed, as measured by the Motor Assessment Scale, grip force, and a decrease in arm muscle spasticity. These results were confirmed by motion data collected by the robot. The results of this study show the feasibility of this robot-assisted therapy with patients presenting a large range of impairment levels. A significant homogeneous improvement in both hand and arm function was observed, which was maintained 6 weeks after end of the therapy.

Motor imagery-based brain–computer interface (MI-BCI) has been proposed as a rehabilitation tool to facilitate motor recovery in stroke. However, the calibration of a BCI system is a time-consuming and fatiguing process for stroke patients, which leaves reduced time for actual therapeutic interaction. Studies have shown that passive movement (PM) (i.e., the execution of a movement by an external agency without any voluntary motions) and motor imagery (MI) (i.e., the mental rehearsal of a movement without any activation of the muscles) induce similar EEG patterns over the motor cortex. Since performing PM is less fatiguing for the patients, this paper investigates the effectiveness of calibrating MI-BCIs from PM for stroke subjects in terms of classification accuracy. For this purpose, a new adaptive algorithm called filter bank data space adaptation (FB-DSA) is proposed. The FB-DSA algorithm linearly transforms the band-pass-filtered MI data such that the distribution difference between the MI and PM data is minimized. The effectiveness of the proposed algorithm is evaluated by an offline study on data collected from 16 healthy subjects and 6 stroke patients. The results show that the proposed FB-DSA algorithm significantly improved the classification accuracies of the PM and MI calibrated models (p < 0.05). According to the obtained classification accuracies, the PM calibrated models that were adapted using the proposed FB-DSA algorithm outperformed the MI calibrated models by an average of 2.3 and 4.5 % for the healthy and stroke subjects respectively. In addition, our results suggest that the disparity between MI and PM could be stronger in the stroke patients compared to the healthy subjects, and there would be thus an increased need to use the proposed FB-DSA algorithm in BCI-based stroke rehabilitation calibrated from PM.

We evaluated the feasibility, safety, and efficacy of a 2D-planar robot for minimally supervised home-based upper-limb therapy for post-stroke hemiparesis. The H-Man, end effector robot, combined with web-based software application for remote tele-monitoring were evaluated at homes of participants. Inclusion criteria were: strokes > 28 days, Fugl-Meyer Motor Assessment (FMA) > 10-60/66, presence of a carer and absence of medical contraindications. Participants performed self-directed, minimally supervised robotics-assisted therapy (RAT) at home for 30 consecutive days, after 2 therapist-supervised clinic on-boarding sessions. Web-based compliance measures were: accessed sessions of > 20 min/day, training minutes/day and active training hours/30 days. Clinical outcomes at weeks 0, 5 (post-training), 12 and 24 (follow-up) consisted of FMA, Action Research Arm Test (ARAT) and WHO-Stroke Specific Quality of Life (SSQOL). To estimate immediate economic benefits of the home-based robotic therapy, we performed cost-effectiveness analysis (CEA), followed by budget impact analysis (BIA). Altogether, all 12 participants completed Home-RAT without adverse events; 9 (75.0%) were males, mean (SD) age, 59.4 years (9.5), median (IQR) stroke duration 38.6 weeks (25.4, 79.6) baseline FMA (0–66) 42.1 ± 13.2, ARAT (0–57) 25.4 ± 19.5, SSQOL (0–245) 185.3 ± 32.8. At week 5 follow-up, mean (SD) accessed days were 26.3 days ± 6.4, active training hours of 35.3 h ± 14.7/30 days, or ~ 6 days/week and 77 training minutes ± 20.9/day were observed. Significant gains were observed from baseline across time; ΔFMA 2.4 at week 5 (FMA 44.5, CI 95% 39.7–49.3, p < 0.05) and ΔFMA 3.7 at week 24 (FMA 45.8, CI 95% 40.5–51, p < 0.05); ΔARAT 2.6 at week 5 (ARAT 28.0, CI 95% 19.3–36.7, p < 0.05), and ΔARAT 4.8 at week 24 (ARAT 30.2, CI 95% 21.2–39.1, p < 0.05). At week 5 follow-up, 91% of participants rated their overall experience as satisfied or very satisfied. Incremental CEA observed savings of -S$144/per cure over 24 weeks, BIA—potentially 12% impact reduction over five years. This study demonstrates the feasibility, acceptability, safety, clinical efficacy, and cost-effectiveness of a home-based, web-enabled telemonitored carer-supervised robotics-aided therapy. Trial registration: NCT05212181 ( https://clinicaltrials.gov ).

Insightful feedback generation for daily home-based stroke rehabilitation is currently unavailable due to the inefficiency of exercise inspection done by therapists. We aim to produce a compact anomaly representation that allows a therapist to pay attention to only a few specific sections in a long exercise session record and boost their efficiency in feedback generation. This study proposes a data-driven technique to model a repetitive exercise using unsupervised phase learning on an artificial neural network and statistical learning on principal component analysis (PCA). After a model is built on a set of normal healthy movements, the model can be used to extract a sequence of anomaly scores from a movement of the same prescription. The method not only works on a standard marker-based motion capture system but also performs well on a more compact and affordable motion capture system based-on Kinect V2 and wrist-worn inertial measurement units that can be used at home. An evaluation of four different exercises shows its potential in separating anomalous movements from normal ones with an average area under the curve (AUC) of 0.9872 even on the compact motion capture system. The proposed processing technique has the potential to help clinicians in providing high-quality feedback for telerehabilitation in a more scalable way.

Proprioception - the sense of body awareness - is frequently impaired after stroke. Clinical tests used to detect proprioceptive impairments require substantial amounts of time and lack sensitivity. In contrast, robotic devices demonstrated that they can objectively assess subjects' proprioceptive acuity. We here present results of a robot-aided bimanual test for wrist proprioception. Ten acute stroke patients with hemiplegic arm weakness and ten healthy subjects, participated to the experiment. Subjects actively moved the wrist of the unaffected limb to match the reference position of the affected wrist, which was passively displaced by the robot to a target across six directions. Our results are as follows. First, healthy individuals showed no effect of hand dominance. No acuity differences were found for between dominant or non-dominant hand. Second, the matching error varied significantly across movement directions. Third, subjects with stroke were less accurate than healthy individuals in matching the wrist position. Fourth, the assessment is clinically feasible. Stroke patients completed the test in about 8 minutes. We conclude that assessment of proprioception in stroke should map positions at different joint degrees of freedom to detect underlying proprioceptive abnormalities.