Listening to sonified electroencephalograms (EEGs) for rhythmic fluctuations is a potentially effective strategy for identifying seizures among at-risk patients presenting with no clinical signs of seizure, according to study findings published in Epilepsia.
Investigators obtained 15-second-long EEG samples (n=84) from patients presenting with altered mental status but no signs of seizure (“silent” seizure) at time of evaluation. A total of 3 EEG experts reviewed the visually presented samples and defined sample patterns as those that were seizures (n=7), those that had seizure-like abnormalities (n=25), or those that featured slowing or normal patterns (n=52).
Medical students (n=34) and nurses (n=30) without EEG interpretation experience received a 4-minute training video and were subsequently asked to determine which sonified EEG audio sample was representative of a seizure or nonseizure. Findings from this survey were compared with reports from neurologists experienced with EEG (n=12) as well as nonexpert medical students (n=29) who reviewed and differentiated the visual EEG samples.
Medical students and nurses who reviewed the sonified EEGs were able to identify seizures with greater sensitivity than EEG experts or nonexperts who visually reviewed the same EEG data (98%±5% and 95%±14% vs 88%±11% and 76%±19%, respectively). Despite these findings, the specificity for detecting seizures (65% and 66%, respectively) was lower than the sensitivity (95% and 98%, respectively) among nonexperienced nurses and students.
In the event where EEGs featured either seizures or seizure-like patterns, nonexpert medical students who auditorily reviewed the EEGs rated samples as seizures with a higher degree of specificity vs experts or nonexpert medical students who reviewed them visually (85%±9% [students] and 82%±12% [nurses] vs 90%±7% [neurologists] and 65%±20% [students]). When visually reviewing the EEG samples, nonexpert medical students demonstrated less sensitivity (76%; P <.001) and specificity (65%; P =.92) for detecting seizures and less sensitivity (62%; P <.001) and specificity (65%; P <.001) for detecting seizures and seizure-like activity.
Considering the investigators selected only single-channel EEGs (eg, T3-T5 or T4-T6) for assessment, it is possible that focal seizures on other channels may have been missed. In addition, findings from this study were obtained in a retrospective fashion and may not be fully representative of findings obtained at the bedside.
The investigators suggest that the use of sonified EEGs “should be used as a triage diagnostic tool to help nonexpert users discern normal or slow activity from seizures and grossly abnormal seizure-like rhythmic or periodic discharges.”
Parvizi J, Gururangan K, Razavi B, Chafe C. Detecting silent seizures by their sound. Epilepsia. 2018;59(4):877-884.