Annual Meeting Abstracts: View

<< Back to Search Results

(Abst. 1.095), 2019

Prospective Evaluation of Generalized Tonic-Clonic Seizure Multimodal Detection: Comparison Between Pediatric and Adult Cohorts in EMU
Authors: Francesco Onorati, Empatica; Giulia Regalia, Empatica; Chiara Caborni, Empatica; William Curt LaFrance, Rhode Island Hospital, Brown University; Andrew S. Blum, Rhode Island Hospital, Brown University; Jonathan Bidwell, Harvard Medical School; Paola De Liso, Bambino Gesù Children’s Hospital; Rima El Atrache, Boston Children’s Hospital, Boston; Tobias Loddenkemper, Boston Children’s Hospital, Boston; Fatemeh Mohammadpour Touserkani, Boston Children’s Hospital, Boston; Rani Sarkis, Brigham and Women's Hospital; Daniel Friedman, New York University Langone Medical Center; Jay Jeschke, New York University Langone Medical Center; Rosalind Picard, Empatica
Content: Rationale: The combination of wrist accelerometer (ACM) and electrodermal activity (EDA) has been shown effective in the detection of generalized tonic-clonic seizures (GTCSs) both in inpatient and outpatient settings (Regalia et al., 2019, Epilepsy Res. 153:79-82). A first multi-center retrospective study reported sensitivity 94.5% and 1 false alarm every 5 days (Onorati et al., 2017, Epilepsia 58(11):1870-1879). FDA requires a prospective study to assess the effectiveness and safety of a seizure detection device, where a predefined detection algorithm (i.e., “fixed and frozen”) is applied on the test cohort. Here we present the results of a prospective inpatient analysis of an ACM- and EDA-based wristband for GTCSs detection (Embrace by Empatica Inc) and investigate performance differences between the pediatric and adult groups. Methods: 141 patients diagnosed with epilepsy were admitted to the EMU at 6 clinical sites: 80 pediatrics (6-21 years) and 61 adults (22-63 years). Overall, 409 days (9,806 hours, 5.727 from pediatrics) of ACM and EDA data were collected. 31 (17 pediatrics) patients experienced a total of 54 (32 pediatrics) GTCSs. Three board-certified neurologists independently marked seizures from v-EEG. The detection algorithm had been previously trained on a dataset from completely different patients. Performances were evaluated in terms of positive percentage agreement (PPA) and false alarm rate per day (FAR). PPA was the ratio of correctly detected GTCSs to total GTCSs, corrected for the presence of multiple GTCSs per patient. The FAR was evaluated two ways: on all of the recordings and on periods of rest, as identified by a proprietary algorithm (Empatica Mate e-diary rest detection algorithm). Results: For pediatrics, the (corrected for multiple-seizures-per-patient) PPA was 91.5% (31 out of 32 GTCSs detected in 17 patients), with a CI of [83.4%-95.3%], not significantly different (p-value>0.25) than the adults' PPA (92.4%, 22 out of 22 GTCSs detected in 14 patients, CI: [91%-93.1%]). The overall FAR on the pediatrics was 1.35 (CI: [0.92-1.87]), significantly higher (p-value<0.01) than adults' (0.67, CI: [0.43-0.95]). The median FAR was 0.24 for pediatrics and 0.17 for adults. The percentage of patients with 0 FA was 44% for pediatrics and 45% for adults. 41 FAs (18 from pediatrics) were actually seizures other than GTCS. Rest periods accounted for 51% of pediatric and 48% of adult total hours. During rest periods, the overall FARs were 0.05 and 0.02 (p-value>0.35) on pediatrics and adults, respectively. Conclusions: Performance complies with and surpasses FDA requirements (lower bound of CI for PPA>0.7 and FAR<2) for both pediatric and adult populations. Detailed inspection of FAR differences between pediatrics and adults showed higher FAR in pediatrics, which is likely due to the fact that children are naturally more active and prone to perform motor, periodic activities. Future work will examine the prospective detection performance and overall impact in outpatient settings. Funding: No funding