Automated Cerebral Edema Detection using Electroencephalography in Post-Cardiac Arrest Patients.

Aim: To develop an electroencephalography (EEG) based machine learning model to identify diffuse cerebral edema in post-cardiac arrest patients and to evaluate its ability to predict edema prior to radiographic detection. Methods: We performed a retrospective, single-center cohort study of adult patients resuscitated from cardiac arrest (2016-2024) who underwent both neuroimaging and EEG monitoring as part of routine clinical care. Machine learning models using Transformer and Long Short-Term Memory architectures were trained to Results: Among 124 patients in the detection model, median age was 53 years, and 74 (59.7%) were male. Sixty-five patients (52.4%) developed diffuse cerebral edema. The best-performing detection model, a Transformer using 4-hour EEG segments, achieved strong performance (AUC 80.0%, accuracy 80.0%, sensitivity 80.0%, specificity 90.0%). In a secondary analysis of 19 patients with diffuse cerebral edema and 19 matched referents, the top-performing prediction model used 8-hour EEG segments (AUC 92.2%, accuracy 90.6%, sensitivity 100%, specificity 88.9%). Conclusion: Diffuse cerebral edema can be identified in survivors of cardiac arrest using machine learning models applied to routine EEG data. In this study, a Transformer based approach demonstrated superior performance for both detection of established edema and identification of EEG patterns that preceded radiographic recognition than LSTM. With validation in larger and independent cohorts, this strategy may enable earlier recognition of evolving cerebral edema during standard EEG monitoring and support timely interventions to mitigate secondary brain injury.