Brain signal complexity and aperiodicity predict human corticospinal excitability

Abstract

BackgroundTranscranial magnetic stimulation (TMS) holds promise for brain modulation with relevant scientific and therapeutic applications, but it is limited by response variability. Targeting state-dependent EEG features such as phase and power shows potential, but uncertainty remains about the suitable brain states. ObjectiveThis study evaluated broadband EEG measures (BEMs), including the aperiodic exponent (AE) and entropy measures (CTW, LZ), as alternatives to band-limited features, such as power and phase, for predicting corticospinal excitability (CSE). MethodsTMS was delivered with randomly applied single pulses targeting the left primary motor cortex in 34 healthy participants while simultaneously recording EEG and EMG signals. Broadband and band-limited EEG features were evaluated for their ability to predict CSE using motor evoked potentials (MEPs) from the right extensor digitorum communis muscle as the outcome measure. ResultsBEMs (AE, CTW) significantly predicted CSE, comparable to beta-band power and phase, the most predictive and spatially specific band-limited markers of motor cortex CSE. Unlike these localized CSE markers at the site of stimulation, BEMs captured more global brain states and greater within-subject variability, indicating sensitivity to dynamic state changes. Notably, CTW was associated with high CSE, while AE was linked to low CSE. ConclusionThis study reveals BEMs as robust predictors of CSE that circumvent challenges of band-limited EEG features, such as narrowband filtering and phase estimation. They may reflect more general markers of brain excitability. With their slower timescale and broader sensitivity, BEMs are promising biomarkers for state-dependent TMS applications, particularly in therapeutic contexts.