Mapping general anesthesia states based on electro-encephalogram transition phases

Abstract

AbstractCortical electro-encephalography (EEG) has become the clinical reference for monitoring unconsciousness during general anesthesia. The current EEG-based monitors classify general anesthesia states simply as underdosed, adequate, or overdosed, with no transition phases among these states, and therefore no predictive power. To address the issue of transition phases, we analyzed EEG signal of isoflurane-induced general anesthesia in mice. We adopted a data-driven approach and utilized signal processing to trackθ- andδ- band dynamics as well as iso-electric suppressions. By combining this approach with machine learning, we developed a fully-automated algorithm. We found that the dampening of theδ-band occurred several minutes before significant iso-electric suppression episodes. Additionally, we observed a distinctγ-frequency oscillation that persisted for several minutes during the recovery phase following isoflurane-induced overdose. Finally, we constructed a map summarizing multiple states and their transitions which can be utilized to predict and prevent overdose during general anesthesia. The transition phases we identified and algorithm we developed may allow clinicians to prevent inadequate anesthesia, and thus individually tailor anesthetic regimens.1Significance statementIn human patients, overdosing during general anesthesia can lead to cognitive impairment. Cortical electro-encephalograms are used to measure the depth of anesthesia. They allow for correction, but not prevention, of overdose. However, data-driven approaches open new possibilities to predict the depth of anesthesia. We established an electro-encephalogram signalprocessing pipeline, and constructed a predictive map representing an ensemble of gradual sedation states during general anesthesia in mice. In particular, we identified key electroencephalogram patterns which anticipate signs of overdose several minutes before they occur. Our results bring a novel paradigm to the medical community, allowing for the development of individually tailored and predictive anesthetic regimens.