Development and External Validation of a Motor Intention-Integrated Prediction Model for Upper Extremity Motor...

Area of Science:

• Neurorehabilitation
• Robotics in Medicine
• Biomedical Engineering

Background:

• Stroke survivors often experience upper extremity (UE) motor deficits, impacting daily living.
• Robotic hand training offers a promising avenue for motor function recovery.
• Predicting meaningful improvements (minimal clinically important differences - MCIDs) is crucial for tailoring rehabilitation strategies.

Purpose of the Study:

• To derive and validate a prediction model for MCID in UE motor function after intention-driven robotic hand training.
• To utilize residual voluntary electromyography (EMG) signals from the affected UE to predict functional gains.
• To identify key predictors for successful motor recovery in chronic stroke survivors.

Main Methods:

• A prospective longitudinal multicenter cohort study involving 131 chronic stroke survivors.
• Collected preintervention data including demographics, clinical scores (Fugl-Meyer Assessment of UE - FMAUE), and EMG-measured motor intention (flexor digitorum, extensor digitorum - ED).
• Employed a classification and regression tree algorithm to predict MCID in FMAUE, using a 1-second MVC-EMG window for ED activity.

Main Results:

• The optimal prediction model incorporated FMAUE (cutoff score 46) and peak ED activity from a 1-second MVC-EMG measurement (AUC, 0.807).
• This model demonstrated superior prediction accuracy compared to models using other EMG time windows or solely clinical scores (AUC, 0.595).
• External validation confirmed the model's robustness (AUC, 0.916), revealing a significant quadratic relationship between ED-EMG and FMAUE increases.

Conclusions:

• A validated prediction model for intention-driven robotic hand training in chronic stroke survivors was established.
• Capturing motor intention via a 1-second EMG window is a significant predictor of MCID in UE motor function post-training.
• High motor intention combined with moderate-to-high or high UE function predicted substantial clinical motor improvement.