Annual Meeting Abstracts: View

<< Back to Search Results

(Abst. 3.161), 2017

Outpatient smartphone videos for classifying epileptic and non-epileptic seizures
Authors: William Tatum, Mayo Clinic - Florida; Lawrence J. Hirsch, Yale University; Robert B. Duckrow, Yale University; David Chen, Baylor College of Medicine Medical Center; Michael Gelfand, Hospital of the University of Pennsylvania; Andrew Blum, Brown University; John Hixson, University of California, San Francisco, CA, USA; William LaFrance, Brown University; Joseph Drazkowski, Mayo Clinic Arizona; Selim R. Benbadis, University of South Florida; and Gregory D. Cascino, Mayo Clinic
Content: Rationale: Video-EEG monitoring (VEM) is the most specific procedure in the evaluation process of patients with suspected seizures, however availability, cost and resource utilization are limited.1,2 A variety of seizure mimics exist that can result in a misdiagnosis and mistreatment. New tools beyond routine E & M are necessary to assist with the diagnosis of paroxysmal neurological disorders to ensure accurate treatment. Home videos and hand-held camcorders are promising surrogates.3,4 Smartphones are a ubiquitous part of a global society with cameras capable of high definition. We sought to determine the usefulness of outpatient smartphone videos in epilepsy (OSmartViE) and report extended data from our preliminary findings of a multi-center prospective study. Methods: A prospective, multi-center, blinded trial of outpatient smartphone video analysis was performed. Patient-generated outpatient smart-phone videos (SV) were acquired and reviewed prior to VEM. Forced choice diagnosis of 1) ES, 2) PNES, or 3) PhysNEE and a corresponding degree of certainty (0-10) was assigned. Epileptologists and senior general neurology residents without special interest in epilepsy were surveyed for a blinded SV diagnosis. Data sharing was performed via HIPPA-protected data transfer utilizing a web-based software application (CaptureProof®). The H&P, SV, and VEM results were obtained using survey forms and compared. Sensitivity, specificity, PPV, NPV was analyzed. Results: Our initial results demonstrated 25 patients [16 F, age 43.33 yrs.; R= 20-80] had H & P, SV and VEM. SV were reviewed in 2.15 mins as opposed to 60 mins with routine H & P and 1443 minutes (3.3 days) with VEM. Most SV had convulsive episodes but 70% were not ES. VEM demonstrated 15/25 (60%) with PNES, 7/25 (28%) with epilepsy, and 3/25 (12%) PhysNEE. The positive and negative predictive value for a SV was good when evaluated by experts and was more predictive than for trainees. No difference in diagnosis of SV with respect to PNES was present. SV quality was adequate for interpretation in nearly 3/4th of cases. Inter-subject differences were present largely based upon technical as opposed to video quality. The primary technical limitation was lack of focus on the area of interest/whole body view. Initial results demonstrated an SV diagnosis had a level of confidence similar to that derived from H & P. At the annual meeting, we intend to present the extended results of our cohort evaluated by 11 epileptologists in our clinical trial. Conclusions: Most SV were recovered from patients with PNES. Epileptologists were more likely to identify non-epilepsy with greater sensitivity than ES with similar specificity. Widespread availability of SV makes them a useful adjunct in epilepsy3-5. We suggest SV are a useful complement to H & P in the outpatient epilepsy clinic and may assist in closing the gap in the diagnosis and triaging patients with PNES. Funding: Funded in part by Mayo Clinic