Abstract
<jats:p>Trait anxiety is a common pre-surgical phenotype and has been linked to adverse postoperative cognitive outcome, but whether it modulates the response to general anaesthesia has not been tested in a controlled design that isolates the anaesthetic from surgical injury. We classified 42 male C57BL/6N mice as low-anxiety (LA, n = 24) or high-anxiety (HA, n = 18) by unsupervised k-medoids clustering on cued fear-retrieval freezing, and exposed all animals to sevoflurane under live-EEG titration without surgery. At adequate anaesthetic depth, HA mice carried a flatter aperiodic (1/f) cortical slope than LA mice (AUC = 0.85, [0.70, 0.95]). Conventional water-cross-maze scoring returned null at post-anaesthetic re-test, but unsupervised decomposition of the same trial videos localised the effect to loss of a specific learned transition in HA mice (AUC = 0.72, [0.56, 0.86]). The dark-phase REMS response reversed direction, with HA mice losing REMS against their own baseline while LA mice gained it (AUC = 0.82, [0.65, 0.94]). The three effects are directionally consistent with modulation of cortical excitation-to-inhibition balance under sevoflurane. The aperiodic exponent and emergence-phase EEG trajectory, both recoverable from routine frontal EEG, are candidate pre-anaesthetic biomarkers.</jats:p>