Dissociating motor learning from recovery in exoskeleton training post-stroke

Area of Science:

• Neurorehabilitation
• Robotics in Medicine
• Motor Recovery Post-Stroke

Background:

• Robotic and gravity-supporting devices are used for upper extremity stroke rehabilitation.
• These devices collect performance data, enabling potential predictive models for motor recovery.
• Patients require time to adapt to exoskeleton devices, learning new forces and visuomotor transformations.

Purpose of the Study:

• To test the hypothesis that performance changes during exoskeleton training involve two distinct phases: a fast learning phase and a slow recovery phase.
• To investigate the relationship between these fast and slow processes in post-stroke arm rehabilitation.

Main Methods:

• Utilized mixed-effect exponential models to analyze movement smoothness during exoskeleton training.
• Collected data from 53 individuals with post-acute stroke over 20 days of training (80 reaching tests).
• Compared stroke patient data with a group of young healthy control subjects.

Main Results:

• Double exponential models better described movement smoothness changes in stroke patients compared to single exponential models.
• Healthy controls showed better fit with single exponential models.
• In stroke patients, improved movement smoothness from the slow component correlated with upper extremity Fugl Meyer score (UEFM) improvement and was influenced by time since stroke.

Conclusions:

• Performance improvements during robotic-assisted training post-stroke involve at least two distinct processes.
• A fast process reflects adaptation to the exoskeleton, while a slow process independently indicates reduction in upper extremity impairment.
• The slow process is more directly linked to actual motor recovery and functional improvement.