The Feasibility of Longitudinal Upper Extremity Motor Function Assessment Using EEG.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Science

Background:

• Objective motor function assessment is vital for stroke recovery evaluation.
• Current methods rely on subjective, questionnaire-based assessments requiring examiner training.
• Electroencephalography (EEG) shows potential for objective motor function scoring, but longitudinal tracking remains unproven.

Purpose of the Study:

• To investigate the feasibility of using EEG data to score motor function longitudinally.
• To evaluate the performance of Convolutional Neural Network (CNN) and Residual Network (ResNet) EEG models in tracking motor function changes over time.
• To validate EEG-based motor function scores against standard clinical assessments in stroke rehabilitation.

Main Methods:

• Previously developed CNN and ResNet EEG models were used to translate EEG data into motor function scores.
• Models were evaluated on a small sample of individuals undergoing a 14-week rehabilitation program.
• Longitudinal performance was assessed by comparing model-derived scores with Fugl-Meyer Assessment (FMA) of the upper extremity.

Main Results:

• Both CNN and ResNet models demonstrated good accuracy and robustness in scoring motor function.
• Average differences were 1.22 points for CNN and 1.03 points for ResNet compared to FMA scores.
• Preliminary evidence supports the use of these EEG-based models for objective, long-term motor function evaluation.

Conclusions:

• EEG-based motor function scoring using CNN and ResNet models is feasible for longitudinal monitoring post-stroke.
• These models offer a promising, objective alternative to traditional assessment methods.
• The findings support the potential application of this technology in clinical rehabilitation settings for tracking recovery.