How reproducible is visual burst detection in ICU EEG? Implications for clinical interpretation and automated detection.

This work was accepted as an abstract for the Arbeitstagung NeuroIntensivMedizin (ANIM 2027), the 44. Jahrestagung der Deutschen Gesellschaft für Neurointensiv- und Notfallmedizin und der Deutschen Schlaganfall-Gesellschaft. This abstract presents research on the reproducibility of visual burst detection in continuous ICU electroencephalography (EEG). It quantifies agreement between expert neurologists, highlights the importance of human inter-rater agreement as a benchmark for automated detection systems. A full manuscript featuring all results and detailed study methods is planned for journal submission. This research was supported by DIZH through the SonNEEM project (“Sonified Neurological EEG Events Monitoring”).