Detecting Connected Consciousness During Propofol-Induced Anesthesia Using EEG Based Brain Decoding

Connected consciousness refers to the state when external stimuli can enter into the stream of our consciousness experience. Emerging evidence suggests that although patients may not respond behaviorally to external stimuli during anesthesia, they may be aware of their surroundings. In this work, we investigated whether EEG based brain decoding could be used for detecting connected consciousness in the absence of behavioral responses during propofol infusion. A total of 14 subjects participated in our experiment. Subjects were asked to discriminate two types of auditory stimuli with a finger press during an ultraslow propofol infusion. We trained an EEG based brain decoding model using data collected in the awakened state using the same auditory stimuli and tested the model on data collected during the propofol infusion. The model provided a correct classification rate (CCR) of 78.74 +/- 11.15% when subjects were able to respond to the stimuli during the propofol infusion. The CCR dropped to 67.29 +/- 8.21% when subjects ceased responding and further decreased to 58.11 +/- 8.15% when we increased the propofol concentration by another 0.2 mu g/ml. After terminating the propofol infusion, we observed that the CCR rebounded to 65.31 +/- 7.11% before the subjects regained consciousness. With the classification results, we provided evidence that loss of consciousness is a gradual process and may progress from full consciousness to connected consciousness and then to disconnected consciousness.