Rapid Electroencephalography and Artificial Intelligence in the Detection and Management of Nonconvulsive Seizures - 18/06/24

Doi : 10.1016/j.annemergmed.2024.04.026

Chase Richard, MD, MBA ^⁎ , David Schriger, MD, MPH, Daniel Weingrow, DO
Division of Emergency Medicine, the University of California Los Angeles, CA

^∗Corresponding Author.

Sous presse. Épreuves corrigées par l'auteur. Disponible en ligne depuis le Tuesday 18 June 2024
Cet article a été publié dans un numéro de la revue, cliquez ici pour y accéder

Abstract

Study objective

Nonconvulsive status epilepticus is a commonly overlooked cause of altered mental status. This study assessed nonconvulsive status epilepticus prevalence in emergency department (ED) patients with acute neurologic presentations using limited electroencephalogram (EEG) coupled with artificial intelligence (AI)-enhanced seizure detection technology. We then compared the accuracy of the AI EEG interpretations to those performed by an epileptologist.

Methods

In a prospective observational cohort analysis, adult patients with unexplained mental status changes identified by emergency physicians received expedited placement of a limited EEG. Data collected encompassed patient demographics, clinical history, EEG interpretations by the AI algorithm and epileptologists, treatments, and disposition determinations.

Results

There were 134 device applications on 132 patients (2 received the device twice) enrolled in the study, but 16 were missing data critical for identification or analysis and 9 did not meet the selection criteria. Of the 108 limited EEGs interpreted by an epileptologist, 69 were abnormal (diffuse slowing, highly epileptiform patterns, or spikes and sharps), 41 were normal, 5 were uninterpretable, and 3 captured episodes of seizure or status epilepticus. Limited EEG AI interpretation detected >90% seizure burden in 2 of 3 cases of seizure or status epilepticus as well as in 2 abnormal EEGs and 1 normal EEG, providing a sensitivity of 66.7% (95% confidence interval 9.4 to 99.2), a specificity of 97.0% (95% confidence interval 91.5 to 99.4), and a disease prevalence of 2.9%.

Conclusion

Limited AI-enhanced EEG can detect nonconvulsive status epilepticus in the ED; however, the technology tended to overestimate seizure burden in our cohort. This study found a lower nonconvulsive status epilepticus prevalence compared to prior literature reports.

Le texte complet de cet article est disponible en PDF.

Plan

Introduction

Background

Importance

Goal of This Investigation

Materials and Methods

Study Design and Setting

Selection of Participants

Characteristics of the Study Subjects

	Please see page XX for the Editor’s Capsule Summary of this article.
	Supervising editor: William J. Meurer, MD, MS. Specific detailed information about possible conflict of interest for individual editors is available at editors.
	Author contributions: CR and DW conceived and designed the study. CR collected the data. The data analysis was performed by CR, DW, and DS. The manuscript was drafted by CR, DW, and DS. CR and DW take responsibility for the paper as a whole.
	Data sharing statement: The entire deidentified data set, data dictionary, and analytic code for this investigation are available on request from the date of article publication by contacting Chase Richard, MD, MBA, by email at crichard@mednet.ucla.edu.
	Authorship: All authors attest to meeting the four ICMJE.org authorship criteria: (1) Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; AND (2) Drafting the work or revising it critically for important intellectual content; AND (3) Final approval of the version to be published; AND (4) Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
	Funding and support: By Annals’ policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article as per ICMJE conflict of interest guidelines (see www.icmje.org/). The authors have stated that no such relationships exist. The authors report this article did not receive any outside funding or support.
	Presentation information: Preliminary data were presented as a poster presentation at the Society for Academic Emergency Medicine Conference in New Orleans, Louisiana on May 12, 2022.