Biophysical modeling of corticospinal tract activation predicts motor contractions in subthalamic deep brain stimulation

Stimulation-induced motor contractions are among the most common dose-limiting side effects in subthalamic nucleus deep brain stimulation for Parkinson's disease, yet no quantitative models exist to predict their occurrence from imaging data. Here, we combine pathway activation modeling of the corticospinal and corticobulbar tracts within the posterior limb of the internal capsule with a data-driven prediction framework. Evaluated by leave-one-patient-out cross-validation across an intraoperative (42 patients, 352 sites) and postoperative sub-cohort (11 patients, 176 contacts), the model explained 31-35% of the variance in observed motor contraction thresholds. The model reliably identified contacts with the lowest and highest motor contraction thresholds within individual electrodes, with strongest performance for distinguishing directional segments at the same electrode level (84% and 64% accuracy; p < 0.001), supporting its potential value for postoperative DBS programming. Back-pro