Identifying Personalized Brain States Predicting Residual Corticospinal Tract Output After Stroke

Study Summary

Transcranial magnetic stimulation (TMS) interventions could feasibly strengthen residual corticospinal tract (CST) connections and promote poststroke hand motor recovery. To maximize the effects of such interventions, they must be delivered during brain activity patterns during which TMS best activates the residual CST and enhances its neural transmission. This approach is termed brain state-dependent TMS. The investigators have recently developed a machine learning framework that identifies personalized brain activity patterns reflecting strong CST activation in neurotypical adults. In this study, the investigators will apply this framework to the poststroke brain for the first time. They will also evaluate relationships between this framework's ability to detect strong and weak CST activation states and measures of CST pathway integrity. Participants will visit the laboratory for two days of testing that are separated by at least one night of sleep. On Day 1, participants will provide their informed consent. The MacArthur Competence Assessment Tool and the Frenchay Aphasia Screening Test will be used to evaluate consent capacity and confirm the presence of expressive aphasia as needed. Afterwards, participants will complete eligibility screening and clinical assessment of upper extremity motor impairment, motor function, and disability using the Upper Extremity Fugl-Meyer Assessment, the Wolf Motor Function Test, and the Modified Rankin Scale. Participants will then be screened for the presence of residual CST connections from the lesioned hemisphere to the affected first dorsal interosseous muscle. Recording electrodes will be attached to this muscle in order to record TMS-evoked twitches in these muscles. During this procedure, single-pulse TMS will be applied to each point of a 1 cm resolution grid covering primary and secondary motor areas of the lesioned hemisphere at maximum stimulator output. If TMS reliably elicits a muscle twitch in the affected first do

Primary Outcome

Personalized machine learning classifiers will be used to discriminate between brain activity patterns during which single-pulse TMS elicits large and small motor-evoked potentials from the affected first dorsal interosseous muscle. After fitting each personalized classifier, F1 values will be calculated and used as performance metrics. F1 values will be compared to the empirical chance level, which will be determined using participant-specific permutation testing.

Conditions Studied

Interventions

• DEVICE Single-pulse TMS

Study Locations (1)