Temporal network instability and low-frequency overconnectivity underlie disorders of consciousness in severe brain injury

Diagnosing Prolonged Disorders of Consciousness (pDoC) is challenging due to the inherent limitations of the behavioral assessments recommended by international rehabilitation guidelines. To address this, we applied temporal graph theory to resting-state EEG recordings from 217 patients with severe Acquired Brain Injury (sABI), with and without pDoC. Higher temporal degree centrality in the θ and δ bands, as well as a higher burstiness coefficient, were found in the pDoC group. In contrast, lower temporal correlation coefficient was observed in the α band in the pDoC group compared to the eMCS group. Furthermore, the analysis of separate frequency bands revealed different behavior of the average topological overlap, with higher values in the eMCS group in the α band and higher values in the pDoC group in the δ band. Thus, repeatability of α-connectivity between t-graphlets, the capability of sustaining more periodic-like changes in all-bands connectivity, and the higher integration of temporal network were found to be correlated with consciousness, providing valuable insights on the time-varying neural architecture of pDoC patients.