Neuronal avalanches as a predictive biomarker for guiding tailored BCI training programs

Motor imagery-based Brain-Computer Interfaces (BCIs) restore control in persons with motor impairments, but up to 30% of users struggle, a phenomenon known as "BCI inefficiency". This study tackles a key limitation of current protocol: the use of fixed-length sessions training paradigms that ignore individual learning variability. We propose a novel approach based on neuronal avalanches, spatiotemporal cascades of brain activities, as biomarkers to characterize and predict user-specific learning. From electroencephalography data across four sessions in 20 subjects, we characterized avalanches by their length and their spatiotemporal size. These features showed significant training and task effects and were found to correlate to BCI performance across sessions. We further assessed their ability to predict BCI success through longitudinal models, achieving up to 91% accuracy, improved by spatial filtering on selected brain regions. These findings demonstrate the utility of neuronal avalanche dynamics as robust biomarkers for BCI training, supporting the development of personalized protocols aimed at mitigating BCI illiteracy.