Manipulating Single-Trial Motor Performance in Chronic Stroke Patients by Closed-Loop Brain State Interaction

Area of Science:

• Neuroscience
• Rehabilitation Medicine
• Motor Control

Background:

• Patients with motor impairments exhibit significant performance variability during repetitive tasks.
• Predicting and optimizing motor task performance on a single-trial basis remains a challenge.

Purpose of the Study:

• To develop a data-driven framework to identify brain states predictive of motor performance.
• To implement a novel gating strategy for optimizing trial initiation based on brain state predictions.
• To evaluate the efficacy of this closed-loop training framework in chronic stroke patients.

Main Methods:

• Extracted oscillatory brain states from pre-trial intervals using a data-driven approach.
• Developed a gating strategy to select trial start times based on predicted performance (good vs. bad).
• Conducted a pilot study with four chronic stroke patients, involving approximately 15 hours of online training.

Main Results:

• Brain states accurately predicted upcoming single-trial motor performance.
• Patients' reaction times were significantly reduced for trials initiated during suitable brain states.
• Two patients demonstrated shorter overall trial durations when training under suitable brain states.

Conclusions:

• The developed brain state-based gating strategy is a viable closed-loop approach for motor training.
• This framework shows potential for enhancing motor rehabilitation in patients with chronic impairments.
• The approach may be transferable to other fields like cognitive rehabilitation and sports science.