AI-BASED ALGORITHM ENCEVIS PERFORMANCE IN THE EPILEPSY MONITORING UNIT: A MULTIFACTORIAL ANALYSIS
DOI:
https://doi.org/10.52340/jecm.2025.06.11Keywords:
EEG, Automated seizure detection, Seizure duration, Seizure localization, AIAbstract
Objective: To evaluate the performance of the ENCEVIS AI-based seizure detection algorithm in long-term EEG recordings and to identify seizure characteristics - specifically localization, rhythmicity, and duration - that influence detection sensitivity.
Methods: This prospective study included 267 prolonged EEG recordings (>3 hours) collected from 2019 to 2023 at Khechinashvili University Hospital. Seizures were visually evaluated by experts, and onset zone, rhythmicity, and duration were defined. ENCEVIS detections were considered true positives if they occurred within 30 seconds before to 60 seconds after expert-defined seizure boundaries. Detection sensitivity was calculated and analyzed using multivariate logistic regression with interaction terms for rhythmicity × duration and localization × rhythmicity.
Results: A total of 114 seizures were identified, of which 65 were correctly detected by ENCEVIS (overall sensitivity: 57.0%). Temporal lobe seizures were detected with the highest sensitivity (71.0%, p < 0.05), while generalized tonic seizures showed the lowest (35.1%). Rhythmic seizures had significantly better detection than arrhythmic seizures (71.4% vs. 22.7%, p < 0.0001). Peak sensitivity was observed for seizures lasting 38–68 seconds. In logistic regression, rhythmicity was the strongest independent predictor of detection (OR = 2.13), with a positive interaction observed between rhythmicity and seizure duration (OR = 1.61). Duration alone and localization alone showed limited predictive value.
Significance: Seizure detection by ENCEVIS is primarily driven by rhythmicity and moderately long duration, particularly in temporal lobe seizures. Arrhythmic, short, and generalized tonic seizures remain underdetected, underscoring the need for algorithmic refinement. Enhancing detection performance for underrepresented seizure types will be essential for AI seizure detection model implementation in clinical practice.
Downloads
References
Tatum WO, Rubboli G, Kaplan PW, et al. Clinical utility of EEG in diagnosing and monitoring epilepsy in adults. Clin Neurophysiol. 2018;129(5):1056-1082. doi:10.1016/j.clinph.2018.01.019
Brogger J, Eichele T, Aanestad E, et al. Visual EEG reviewing times with SCORE EEG. Clin Neurophysiol Pract. 2018;3:59-64. doi:10.1016/j.cnp.2018.03.002
Benbadis SR, Lin K. Errors in EEG interpretation and misdiagnosis of epilepsy: which EEG patterns are overread? Eur Neurol. 2008;59(5):267-71. doi: 10.1159/000115641.
Reus, E. E. M., Visser, G. H., van Dijk, J. G., & Cox, F. M. E. (2022). Automated seizure detection in an EMU setting: Are software packages ready for implementation? Seizure, 96, 13-17. https://doi.org/10.1016/j.seizure.2022.01.009
Koren, J., Hafner, S., Feigl, M., Baumgartner, C. (2021). Systematic analysis and comparison of commercial seizure-detection software. Epilepsia, 62, 426-438. https://doi.org/10.1111/epi.16812
Fürbass, F., Kampusch, S., Kaniusas, E., Koren, J., Pirker, S., Hopfengärtner, R., Stefan, H., Kluge, T., & Baumgartner, C. (2017). Automatic multimodal detection for long-term seizure documentation in epilepsy. Clinical Neurophysiology, 128, 1466-1472. https://doi.org/10.1016/j.clinph.2017.05.013
Fürbass F, Kampusch S, Kaniusas E, et al. Automatic multimodal detection for long-term seizure documentation in epilepsy. Clin Neurophysiol. 2017;128:1466-1472. doi:10.1016/j.clinph.2017.05.013
Tsereteli A, Okujava N, Malashkhia N, Liluashvili K, Weerd A. The ENCEVIS algorithm in the EMU and the factors affecting its performance: our experience. Epilepsy Behav Rep. 2024;26:100656. doi:10.1016/j.ebr.2024.100656
Fürbass F, Ossenblok P, Hartmann M, et al. Prospective multi-center study of an automatic online seizure detection system for epilepsy monitoring units. Clin Neurophysiol. 2015;126(6):1124-1131. doi:10.1016/j.clinph.2014.09.023
Downloads
Published
How to Cite
Issue
Section
