T23. Individuals without EEG training can detect subclinical seizures and seizure-like activity by listening to sonified EEG

Introduction The traditional approach to interpreting EEG requires physicians with formal training to visually assess the waveforms. This approach can be less practical in critical settings where a trained EEG specialist may not be readily available to diagnose subclinical seizures, such as non-convulsive status epilepticus, in patients with altered mental status. As a proof of concept, we designed the current study to explore whether individuals without EEG training can detect ongoing seizures by simply listening to one channel of sonified EEG instead of viewing 20+ channels of EEG. Methods We selected 84, 15-s long, EEG samples to represent various conditions commonly seen in the ICU. Our reference standard was defined by unanimous agreement of 3 epileptologists reviewing visually displayed EEGs in double banana montage. Samples were classified as seizure (focal or generalized, n = 7), seizure-like (LPD, GPD, or burst suppression, n = 25), or non-periodic non-rhythmic (normal or focal/generalized slowing, n = 52). EEG data were then converted to sound files (separating left and right hemispheres to 168 clips, also 15 s long) using a novel sonification method. After a brief training video, medical students (n = 34) and nurses (n = 30) were asked to indicate each audio sample as “seizure” or “non-seizure”. We then compared their performance (sensitivity and specificity; reported as mean ± SD) with that of EEG experts [epilepsy attendings with >10 years of experience (n = 2) and epilepsy fellows (n = 7)] and some of the medical students (n = 29) who also diagnosed the same EEGs on visual display. Our prediction was that students and nurses without EEG training would be able to detect seizures in sonified EEGs with high sensitivity while they will be able to differentiate non-seizure or non-seizure-like events from seizure or seizure-like events with high specificity . Results Non-experts listening to single-channel sonified EEGs detected seizures with remarkable sensitivity (students: 98 ± 5%; nurses: 95 ± 14%) compared to experts or non-experts reviewing the same EEGs on visual display (attendings: 100%; fellows: 90 ± 11%; students: 76 ± 19%). If the EEGs contained seizures or seizure-like activity, non-experts listening to sonified EEGs rated them as seizures with high specificity (students: 85 ± 9%; nurses: 82 ± 12%) compared to experts or non-experts viewing the EEGs visually (attendings: 95 ± 1%; fellows: 91 ± 7%; students: 65 ± 20%). Conclusion Our study is the first of its kind to confirm that individuals without EEG training, such as medical students and nurses, can detect ongoing seizures or seizure-like rhythmic periodic activity by merely listening to short duration of sonified EEG. While sonification of EEG cannot replace the traditional approaches to EEG interpretation, it provides a meaningful triage tool for fast assessment of patients with suspected subclinical seizures.