Cerebral Hemodynamics Modulation in Subacute Stroke: Predictors of Responsiveness to Common Peroneal NMES

Background: Enhancing cerebral perfusion after stroke may aid recovery, but individual responsiveness to neuromuscular electrical stimulation (NMES) remains unclear. Objective: To identify predictors of cerebral hemodynamic response to common peroneal NMES in subacute stroke. Methods: In this observational study (RETRAIN Phase 1, NCT06614400), 18 patients underwent functional near-infrared spectroscopy (fNIRS) during NMES at varying postures and intensities. Responders were defined by top-quartile increases in oxygenated hemoglobin (HbO). Predictors were assessed using AUROC screening and LASSO logistic regression with crossvalidation. Results: The final model (AUROC = 0.723) retained 13 predictors, including cortical lesion location (OR = 1.66), atrial fibrillation (OR = 1.28), and ASPECTS thalamic and caudate involvement. Permutation analysis confirmed predictor robustness. Conclusion: This study demonstrates a clinically interpretable machine learning framework to stratify stroke patients by NMES responsiveness. Larger studies with outcome validation are warranted.