Predicting Robot-Based Tracking Performance After Stroke Using Lesion Features and Clinical Assessments: A Pilot Study

Clinical assessments in 18 chronic stroke participants (70+/-6 years) significantly predicted robot tracking error (R^2=0.45) and delay (R^2=0.22), outperforming lesion features.

• Why it matters: Personalized post-stroke rehabilitation needs efficient prediction of functional outcomes, but the added value of imaging-derived lesion metrics over standard clinical testing is unclear.
• What they did: Participants completed a robot-based tracking task plus motor and cognitive batteries; elastic net regression compared clinical-only, lesion/chart-only, and combined models using leave-one-subject-out cross-validation, bootstrapping, and permutation testing with false discovery rate correction.
• The result: The combined model significantly predicted tracking error (R^2=0.37), delay (R^2=0.19), and velocity error response (R^2=0.11), while the imaging model predicted none; Box and Blocks Test and Color Trails 1 were stable, and the right dorsal attention network was a consistent predictor in several combined models.

AI summary of the abstract · may contain errors · not medical advice · read the paper.