Motor imagery interventions in stroke patients modulate their brain-heart network dynamics

Background Motor imagery is widely used in post-stroke rehabilitation research, as it is thought to promote neuroplastic mechanisms underlying motor recovery. The characterization of the associated neurophysiological patterns is therefore crucial to improve our understanding and eventually optimize rehabilitative brain-computer interface applications. Emerging evidence suggests that cardiac dynamics are actively related to motor-related brain processes, yet the role of brain-heart interactions (BHI) in post-stroke recovery remains largely unexplored. This study aimed to investigate how BHI evolve during motor imagery interventions in stroke patients, with the hypothesis that BHI would exhibit a progressive change across motor imagery interventions. Methods We analyzed two independent cohorts comprising a total of 15 stroke survivors with upper or lower limb motor impairments undergoing motor imagery-based interventions. BHI were quantified by assessing the coupling between cardiac sympathetic activity and EEG-derived brain network metrics, including clustering and assortativity, across multiple sessions. We examined BHI in three different contexts: i) when motor imagery was combined with electrical stimulation, ii) throughout longitudinal motor imagery interventions, and iii) across longitudinal assessments of behavioral changes. Results Motor imagery combined with electrical stimulation on the affected limb modulated BHI across a wide range of frequencies. In longitudinal interventions, BHI showed a progressive increase, specifically reflected in stronger coupling between the cardiac-sympathetic index (CSI) and beta-band clustering. Moreover, changes in functional independence were tracked by variations in the coupling between CSI and EEG gamma-band networks assortativity. Conclusions Our findings suggest that BHI dynamically reorganize during post-stroke recovery and are modulated by motor imagery interventions. The observed change in BHI suggests that these interactions may reflect underlying neuroplastic processes. As such, BHI represent a promising class of multisystem biomarkers for tracking recovery and could inform the development of more adaptive and physiologically grounded BCI-based rehabilitation strategies.