Predicting response to non-invasive brain stimulation in post-stroke upper extremity motor impairment: the importance of neurophysiological and clinical biomarkers.

Abstract

Background: Non-invasive brain stimulation (NIBS) is a promising approach to enhance upper extremity motor impairment (UEMI) recovery in post-stroke individuals. However, variability in treatment response poses a significant challenge. Identifying neurophysiological and clinical biomarkers that predict NIBS response could improve personalization and treatment efficacy.

Objectives: This study aims to determine the predictive relevance of neurophysiological and clinical biomarkers for responses to NIBS in post-stroke UEMI using a machine learning model.

Methods: This secondary analysis involved 63 post-stroke individuals with UEMI (age 56.9 ± 11.1 years). A support vector machine model was used to assess the importance of two neurophysiological biomarkers-brain activity in the lesioned hemisphere quantified using quantitative electroencephalography (power ratio index, PRI) and corticospinal tract (CST) integrity assessed via transcranial magnetic stimulation-and one clinical biomarker-the level of UEMI assessed with Fugl-Meyer upper extremity (FMA-UE)-in predicting responders (ΔFMA-UE ≥ 5 points) and those with excellent response (ΔFMA-UE ≥ 10 points) to NIBS based on the change of FMA-UE before and after treatment.

Results: Of the 63 participants, 42 (65%) were classified as responders, and 14 (22%) demonstrated excellent responses. Predictive importance for responders was 0.78 for PRI-LH, 0.21 for UEMI level, and 0.01 for CST integrity. For predicting excellent responses, PRI-LH had an importance of 0.39, UEMI level 0.37, and CST integrity 0.24.

Conclusions: The study highlights the importance of electrical brain activity in the LH and UEMI level in predicting NIBS responders and excellent responses, with CST integrity being particularly valuable for excellent outcomes.