Clinical efficacy and cortical network modulation of a multimodal sensory feedback upper-limb rehabilitation robot for post-stroke upper-limb recovery: a randomized controlled trial.
Zhao L, Zhang J, et al. • Frontiers in neurology • 2026
Adding multimodal sensory-feedback robot-assisted training to dose-matched conventional rehabilitation may improve short-term post-stroke upper-limb motor recovery, with fNIRS revealing intervention-related changes in cortical activation and resting-state functional connectivity.
Key Findings
Results
The experimental group showed significant improvement in FMA-UE after intervention, whereas the control group showed a smaller, non-significant improvement, with the magnitude of pre-post change differing significantly between groups.
38 stroke patients were randomly assigned to control (n=19) or experimental (n=19) groups
No significant between-group difference in FMA-UE was observed at baseline (p > 0.05)
The between-group difference in pre-post FMA-UE change was statistically significant (p < 0.05)
The control group received conventional rehabilitation; the experimental group received dose-matched conventional rehabilitation plus multimodal sensory-feedback robot-assisted training
Results
Post-intervention resting-state fNIRS analyses showed enhanced fronto-motor functional connectivity in the experimental group compared to the control group.
Resting-state functional connectivity was measured using Fisher z-transformed connectivity (zFC)
No significant between-group differences in zFC were observed at baseline
Post-intervention analyses revealed enhanced fronto-motor zFC in the experimental group
Enhanced connections involved PreM&SMC (premotor and supplementary motor cortex), DLPFC (dorsolateral prefrontal cortex), and FEF (frontal eye fields)
Results
Task-evoked fNIRS revealed channel-specific group × time interaction effects in HbO responses during shoulder, elbow, and finger flexion tasks.
fNIRS was used to evaluate task-evoked cortical activation (HbO) during three motor tasks: shoulder, elbow, and finger flexion
Significant group × time interaction effects were found at selected channels for each of the three tasks
Post-hoc analyses showed significant post-intervention reductions in cortical activation at selected channels in the control group
The experimental group showed a relatively preserved cortical activation pattern after intervention
Methods
Multimodal sensory-feedback robot-assisted training was added as a supplement to conventional rehabilitation in a dose-matched design to evaluate its incremental benefit.
The study used a randomized controlled trial design registered under ChiCTR2400080101
The experimental group received conventional rehabilitation plus multimodal sensory-feedback robot-assisted training, with total dose matched to the control group
Assessments were conducted before and after the intervention period
Both clinical outcomes (FMA-UE) and neurophysiological outcomes (fNIRS resting-state and task-evoked) were measured
Methods
The study used functional near-infrared spectroscopy (fNIRS) to characterize both resting-state functional connectivity and task-evoked cortical activation as neurophysiological markers of rehabilitation-related brain changes.
Task-evoked activation was indexed by HbO (oxygenated hemoglobin) changes during motor tasks
Three motor tasks were evaluated: shoulder flexion, elbow flexion, and finger flexion
fNIRS was used as a non-invasive neuroimaging tool to evaluate cortical network modulation associated with the intervention
What This Means
This research suggests that adding a robot-assisted training system with multimodal sensory feedback (combining multiple types of sensory information such as visual, auditory, or tactile cues) to standard stroke rehabilitation can lead to greater improvements in arm and hand movement than conventional therapy alone. In a randomized controlled trial of 38 stroke patients, those who received the robot-assisted training in addition to their usual rehabilitation showed significantly larger gains on a standard arm function test (the Fugl-Meyer Assessment for the Upper Extremity) compared to patients who received only conventional rehabilitation.
Beyond the physical improvements, the study used a brain imaging technique called functional near-infrared spectroscopy (fNIRS) to look at what was happening in the brain during rest and during movement tasks. Patients in the robot-assisted group showed strengthened connections between frontal and motor brain regions at rest after treatment, and their brains maintained relatively stable patterns of activity during arm movement tasks. In contrast, patients in the conventional therapy group showed reductions in brain activation during movement tasks after treatment, which the researchers interpreted as a less favorable neural pattern.
This research suggests that combining sensory-rich, robot-assisted training with conventional stroke rehabilitation may not only improve physical arm function in the short term but may also promote beneficial reorganization of brain networks involved in movement planning and execution. The fNIRS findings provide preliminary evidence of the neurological mechanisms behind these improvements, though larger and longer-term studies will be needed to confirm these results and determine how lasting the benefits are.
Zhao L, Zhang J, Wu J, Cao Y, Lu Z, Hua X, et al.. (2026). Clinical efficacy and cortical network modulation of a multimodal sensory feedback upper-limb rehabilitation robot for post-stroke upper-limb recovery: a randomized controlled trial.. Frontiers in neurology. https://doi.org/10.3389/fneur.2026.1820579