Towards real-time personalized brain state-dependent TMS to enhance poststroke hand rehabilitation

PROJECT SUMMARY Stroke commonly disrupts the corticospinal tract (CST) and impairs hand function, but current rehabilitation approaches incompletely restore pre-stroke hand motor control. Transcranial magnetic stimulation (TMS) interventions that target the residual CST and strengthen its neural transmission are promising candidates for

enhancing paretic hand muscle activation during rehabilitation and promoting true hand motor recovery. To maximize their therapeutic effects, poststroke TMS interventions must reliably activate the residual CST and enhance residual CST transmission. We and others recently showed that neurotypical adults exhibit

spontaneous resting brain activity patterns during which TMS most strongly activates the CST (i.e., strong CST states). When TMS interventions are delivered during these strong CST states, they preferentially enhance CST transmission and enhance motor learning. However, virtually all poststroke TMS interventions are uncoupled

from strong CST states, such that only a fraction of TMS stimuli occur when the residual CST is best activated by TMS and neural transmission within it is most likely to be enhanced. Instead, poststroke TMS interventions should be delivered solely during strong CST states. Because each stroke survivor has a unique pattern of brain

damage, recovery-related brain reorganization, and motor impairment, these strong CST states must be fully personalized. To address these issues, we developed a novel real-time EEG-informed TMS system that delivers stimulation during personalized strong CST states. Our system uses personalized machine learning classifiers

and participant-specific datasets to identify multivariate EEG activity patterns that predict strong CST activation, indexed by large motor-evoked potentials (MEPs). Once these EEG activity patterns are detected in real-time, TMS is delivered. In this project, we will demonstrate the feasibility of real-time EEG-informed personalized brain

state-dependent TMS in chronic stroke survivors while also determining hand impairment inclusion criteria for future clinical trials that investigate this promising personalized brain stimulation approach. In Aim 1, we will establish feasibility of real-time personalized brain state-dependent TMS in the poststroke CST system by

comparing MEP amplitudes elicited during strong CST states to those elicited during random CST states. In Aim 2, we will correlate differences in MEP amplitudes elicited during strong versus random CST states with poststroke hand impairment. We will then identify the hand impairment levels of stroke survivors who exhibit

larger MEPs during personalized strong than random CST states; previous literature suggests that these stroke survivors are most likely to benefit from future personalized poststroke brain state-dependent TMS interventions. Overall, this project will demonstrate feasibility of personalized poststroke brain state-dependent TMS and

establish hand impairment inclusion criteria for future clinical trials that test this novel approach. Our results are expected to motivate an R01 proposal that investigates the therapeutic effects of personalized poststroke brain state-dependent TMS interventions.