Postoperative delirium (POD) is a common and severe complication among older surgical patients. Perioperative electroencephalography (EEG) monitoring provides a noninvasive, real-time window into cerebral neural activity. Traditionally, EEG has been simplified into single indices, such as the bispectral index (BIS), to guide anesthetic depth titration. However, recent large-scale clinical trials have demonstrated that avoiding EEG suppression based solely on such indices does not reduce the incidence of POD. This discrepancy partly arises because conventional indices are primarily derived from frontal EEG signals and fail to fully exploit the multidimensional information distributed across the whole brain. This review highlights that high-dimensional features embedded in raw EEG—including spectral characteristics, oscillatory dynamics, brain network connectivity, and aperiodic components—serve as key biomarkers for identifying individual neural vulnerability and POD risk. We systematically summarize spatiotemporal EEG phenotypes associated with POD pathophysiology, encompassing age-related baseline features, intraoperative high-risk patterns (such as burst suppression), and abnormal emergence trajectories indicative of failed brain network reorganization. Accumulating evidence indicates that aberrations in these EEG features are closely associated with POD occurrence and adverse clinical outcomes. Therefore, future EEG interpretation must move beyond single “depth” indices toward a mechanism-driven assessment that integrates high-dimensional EEG features with individual baseline characteristics, thereby providing novel strategies for the precise prevention of POD.
Tuerhong et al. (Sun,) studied this question.