Fig. 1: Analysis pipeline.
From: Detection of EEG dynamic complex patterns in disorders of consciousness
A Offline calculation of brain states: We utilized three datasets from different centers, comprising healthy controls and three patient categories (Minimally Conscious Syndrome [MCS], Unresponsive Wakefulness Syndrome[UWS], and Acute patients). Windowed wSMI matrices were computed from EEG data, followed by clustering analysis to identify 5 distinct brain states. The probability and association with patient prognosis were then evaluated. B Real-time calculation of brain states: Simulating a bedside scenario, we processed 16 s of raw EEG data every 24 s to generate raw-data wSMI matrices. By matching these matrices to the pre-defined brain states obtained offline, we established real-time brain state identification.
