Robustness of individualized inferences from longitudinal resting state dynamics

Abstract

Individual human brains plastically reorganize over extended timescales ranging from days to years, which makes these changes difficult to track. One promising tracking indicator is the neural activity at rest. However, the rest state (RS) is vulnerable to incidental inter-day cognitive variability that can confound the detection of changes to underlying neurophysiology. Here we show that this confounding is minimized by tracking changes to the distinctiveness of an individual’s RS activity, which is shaped by individual neurophysiology. Using longitudinal (5-day) RS acquired with EEG, we devised empirical simulations of distinctiveness changes confounded by cognitive variation. These inter-day changes were correctly classified with over 96% accuracy from 2-second snapshots of instantaneous oscillatory power from RS and confounded-RS. Surprisingly, the individual indicators of distinctiveness were concentrated at characteristic fronto-central and occipital zones. These findings support the suitability of longitudinal RS for individualized inferences about neurophysiological change in health and disease.