Predicting interindividual response to theta burst stimulation in the lower limb motor cortex using machine learning

Area of Science:

• Neuroscience
• Neuromodulation
• Machine Learning in Medicine

Background:

• Theta burst stimulation (TBS) is crucial for inducing neural plasticity in neurological disorder treatments.
• Variability in TBS-induced synaptic plasticity in the primary motor cortex limits clinical use.
• Predictive modeling requires exploring factors influencing TBS variability.

Purpose of the Study:

• To identify factors predicting variability in TBS-induced synaptic plasticity in the lower limb motor cortex.
• To apply machine learning for uncovering novel predictive factors.
• To compare predictors for intermittent (iTBS) and continuous (cTBS) TBS.

Main Methods:

• Utilized a prior dataset (Katagiri et al., 2020).
• Employed machine learning algorithms to analyze neurophysiological factors.
• Validated predictive models using metrics like AUC, accuracy, precision, recall, and F1 scores.

Main Results:

• Machine learning models achieved significant predictive performance (AUCs of 0.85 for iTBS, 0.69 for cTBS).
• Key predictors identified: motor evoked potential amplitude for iTBS and intracortical facilitation for cTBS.
• Models demonstrated good accuracy, precision, recall, and F1 scores for both iTBS and cTBS.

Conclusions:

• Baseline neurophysiological factors can predict TBS effects variability.
• Machine learning enhances the understanding of non-linear changes in synaptic plasticity.
• Findings offer insights for personalized TBS application in clinical settings.