Sensor-assisted soft robotic glove training did not demonstrate superiority over supervised GRASP for upper limb function post-stroke, though exploratory subgroup analyses identified a signal suggesting improved distal hand motor recovery in participants with greater baseline impairment with sufficient therapy exposure.
Key Findings
Results
Both the soft robotic glove (SSR) and supervised GRASP groups showed improvement across primary and secondary outcomes, with no statistically significant between-group differences in the primary analysis.
61 participants were randomised across 2 inpatient rehabilitation units between July 2023 and May 2025
Modified intention-to-treat population was 58 (SSR n=28, control n=30) after 3 participants did not complete T2 assessment
Primary outcome was Fugl-Meyer Assessment of Upper Extremity (FMA-UE), assessed at baseline (T1), post-intervention (T2), and 3-month follow-up (T3)
No statistically significant differences were observed between groups on the primary or any secondary outcomes
The trial was underpowered as recruitment ceased before the planned sample size was reached due to funding constraints
Results
In a post-hoc exploratory subgroup analysis, the SSR group showed a statistically significant improvement in FMA-UE distal sub-score at post-intervention compared to the control group among participants with greater baseline impairment and sufficient therapy exposure.
Subgroup criteria were baseline FMA-UE < 35 and completion of ≥10 sessions (n=38)
Mean difference in FMA-UE distal sub-score at T2 was 3.06 (95% CI 0.02 to 6.09, p=0.049) favouring the SSR group
The statistically significant difference in FMA-UE distal sub-score was not maintained at 3-month follow-up (T3)
This analysis was exploratory and post-hoc, meaning it was not a pre-specified primary or secondary analysis
Results
A numerically higher proportion of SSR group participants achieved the minimal clinically important difference (MCID) for Stroke Impact Scale-16 (SIS-16) at post-intervention, but this did not reach statistical significance.
77.8% of SSR participants vs. 55.0% of control participants achieved MCID for SIS-16 at T2
Adjusted odds ratio was 6.52 (95% CI 1.00 to 42.58, p=0.050)
Absolute risk difference was 22.8% (95% CI -7.2% to 47.4%)
The difference did not reach the conventional threshold for statistical significance
Methods
The trial enrolled adults aged 21–90 years with ischaemic or haemorrhagic stroke and unilateral upper-limb impairment, randomised to SSR or supervised GRASP training for up to 15 sessions over 3 weeks.
323 individuals were screened and 61 were randomised
Intervention consisted of up to 15 sessions, 30–45 minutes per session, 5 days per week for up to 3 weeks
The study was assessor-blinded and conducted in 2 inpatient rehabilitation units
Secondary outcomes included Stroke Upper Limb Capacity Scale, DurüÖz Hand Index, and Stroke Impact Scale-16
The comparator was the Graded Repetitive Arm Supplementary Programme (GRASP), a supervised structured upper-limb exercise programme
Discussion
The trial was underpowered due to early cessation of recruitment before the planned sample size was reached.
Recruitment ceased before the planned sample size was reached due to funding constraints
The authors explicitly stated the trial was 'underpowered to detect the prespecified between-group difference'
61 of the planned number of participants were randomised
This limitation affects the interpretation of null findings between the two groups
Background
Soft robotic devices were described as potentially useful adjuncts for stroke rehabilitation due to their ability to provide higher intensity training with greater comfort and portability compared to rigid exoskeletons.
The rationale for SSR use included allowing higher intensity training while maintaining comfort and portability
The study used a sensor-assisted soft robotic glove (SSR) specifically designed for upper-limb rehabilitation
The comparator GRASP was chosen as a supervised active training control
The study context was subacute stroke inpatient rehabilitation
What This Means
This research suggests that using a sensor-assisted soft robotic glove for upper-limb rehabilitation after stroke does not produce significantly better arm and hand recovery than a structured supervised exercise programme called GRASP, at least when looking at the overall group of patients studied. Both groups improved over the course of their rehabilitation, but there was no clear winner between the two approaches on the main measures of arm function. The study was conducted in inpatient rehabilitation units with 61 stroke survivors who had weakness in one arm, and participants received up to 15 sessions of their assigned therapy over three weeks.
When the researchers looked more closely at a specific subgroup—patients who had more severe arm weakness at the start and who completed at least 10 therapy sessions—they found a signal that the robotic glove group had better recovery of hand and finger movement immediately after the treatment period ended. However, this difference did not persist at a 3-month follow-up, and this analysis was exploratory rather than pre-planned, which means it should be interpreted cautiously. A higher proportion of robotic glove users also reported improvements in how stroke was affecting their daily life, though this difference just missed the threshold for statistical significance.
An important limitation of this study is that it enrolled fewer participants than originally planned because funding ran out, making it less able to detect real differences between the two treatments. This research suggests that soft robotic gloves may hold promise particularly for patients with more severe hand weakness who receive an adequate dose of therapy, but larger and better-powered studies are needed to confirm whether this technology offers meaningful advantages over established supervised exercise programmes in stroke rehabilitation.
Tay K, Tam P, Xie Q, Lai H, Yuen W, Lim X, et al.. (2026). Efficacy of a soft robotic exoskeleton for upper-limb rehabilitation in subacute stroke inpatient rehabilitation: a randomised controlled trial.. Journal of neuroengineering and rehabilitation. https://doi.org/10.1186/s12984-026-02133-0