An automated EEG algorithm detect polymorphic delta activity in acute encephalopathy presenting as postoperative delirium

Abstract Aim Delirium, a clinical manifestation of acute encephalopathy, is often unrecognised. An important electroencephalography (EEG) characteristic of acute encephalopathy is polymorphic delta activity (PDA), which can be detected automatically. We aimed to study whether automated assessment of PDA in unselected EEG could detect acute encephalopathy that presents clinically as delirium. Methods We assessed PDA in 145 elderly patients using the first 96 seconds of unselected single-channel EEG (Fp2-Pz). We compared fully automated PDA detection with visual inspection by EEG experts. Additionally, we tested its performance as a delirium monitor by comparing PDA detection with a standardized delirium assessment by a clinical expert panel. Results PDA detection showed an area under the receiver operating characteristic (AUC) of 0.86 (95%CI 0.81–0.90) compared to EEG experts. When compared with the delirium classification of clinical experts, PDA detection achieved an AUC of 0.78 (95%CI 0.71–0.85). PDA detection correlated with the likelihood of delirium, its severity and the levels of attention and consciousness (all p<0.001). Conclusion Automated PDA detection in unselected, single-channel EEG can classify acute encephalopathy clinically presenting as delirium. Significance A fully automated EEG algorithm can assist in the recognition of delirium.