Abstract
Background:
Depth of anesthesia is routinely monitored using electroencephalographic (EEG) indices such as the Bispectral Index (BIS), which reduce cortical activity to a single numerical real-time series. However, these scalar descriptors fail to capture inter-individual variability, drug-specific EEG signatures, and transitions between brain states.
Methods:
We analyzed transient EEG patterns during general anesthesia induced by propofol and sevoflurane in 89 patients. Iso-electric suppressions (IES), $\alpha$-suppressions, and spectral power in the $\alpha$ and $\delta$ bands were quantified across controlled concentration plateaus. We compared conventional BIS monitoring with a novel State-Chart representation mapping EEG activity into discrete depth states.
Results:
Increasing hypnotic concentration produced a consistent shift from $\alpha$-dominant to $\delta$-dominant activity and a reduction in maximal $\alpha$ frequency. IES occurred predominantly at high propofol concentrations and remained infrequent under sevoflurane. Compared with BIS, the State-Chart showed a marked reduction in temporal variability (0.5 $\pm$ 0.6 vs 4.0 $\pm$ 2.9; $p<10^{-90}$). Importantly, our State-Chart separates stable from transition regimes and discriminates identical concentration levels. Statistics computed over prior dosing (ascending vs descending), reveals refined clustering not captured by BIS.
Conclusions:
Anesthesia depth drugs concentrations vs EEG effects is more accurately described by structured EEG state transitions than by a single numerical index. The State-Chart provides a stable, physiologically interpretable representation of brain activity and reveals transition states preceding deep cortical suppression, supporting individualized and predictive monitoring of anesthesia.