Daniel Jacobs1, Yuhan H Liu2, Trevor Hilton1, Martin Del Campo3, Peter L Carlen1,3,4, Berj L Bardakjian1,2. 1. 1Institute of Biomaterials and Biomedical Engineering, University of TorontoTorontoONM5S 3G9Canada. 2. 2Department of Electrical and Computer EngineeringUniversity of TorontoTorontoONM5S 3G9Canada. 3. 3Department of Medicine (Neurology)Toronto Western HospitalTorontoONM5T 2S8Canada. 4. 4Krembil Research Institute, University Health NetworkTorontoONM5T 2S8Canada.
Abstract
OBJECTIVE: To investigate the feasibility of improving the performance of an EEG-based multistate classifier (MSC) previously proposed by our group. RESULTS: Using the random forest (RF) classifiers on the previously reported dataset of patients, but with three improvements to classification logic, the specificity of our alarm algorithm improves from 82.4% to 92.0%, and sensitivity from 87.9% to 95.2%. DISCUSSION: The MSC could be a useful approach for seizure-monitoring both in the clinic and at home. METHODS: Three improvements to the MSC are described. Firstly, an additional check using RF outputs is made prior to alarm to confirm increasing probability of a seizure onset state. Secondly, a post-alarm detection horizon that accounts for the seizure state duration is implemented. Thirdly, the alarm decision window is kept constant.
OBJECTIVE: To investigate the feasibility of improving the performance of an EEG-based multistate classifier (MSC) previously proposed by our group. RESULTS: Using the random forest (RF) classifiers on the previously reported dataset of patients, but with three improvements to classification logic, the specificity of our alarm algorithm improves from 82.4% to 92.0%, and sensitivity from 87.9% to 95.2%. DISCUSSION: The MSC could be a useful approach for seizure-monitoring both in the clinic and at home. METHODS: Three improvements to the MSC are described. Firstly, an additional check using RF outputs is made prior to alarm to confirm increasing probability of a seizure onset state. Secondly, a post-alarm detection horizon that accounts for the seizure state duration is implemented. Thirdly, the alarm decision window is kept constant.
Authors: Prisca R Bauer; Roland D Thijs; Robert J Lamberts; Demetrios N Velis; Gerhard H Visser; Else A Tolner; Josemir W Sander; Fernando H Lopes da Silva; Stiliyan N Kalitzin Journal: Brain Date: 2017-03-01 Impact factor: 13.501
Authors: Chunsheng Li; Daniel Jacobs; Trevor Hilton; Martin Del Campo; Yotin Chinvarun; Peter L Carlen; Berj L Bardakjian Journal: IEEE Trans Biomed Eng Date: 2016-12 Impact factor: 4.538
Authors: Sinisa Colic; Robert G Wither; Min Lang; Liang Zhang; James H Eubanks; Berj L Bardakjian Journal: J Neural Eng Date: 2016-11-30 Impact factor: 5.379
Authors: Stefan Beyenburg; Alex J Mitchell; Dieter Schmidt; Christian E Elger; Markus Reuber Journal: Epilepsy Behav Date: 2005-09 Impact factor: 2.937
Authors: Mirna Guirgis; Yotin Chinvarun; Martin Del Campo; Peter L Carlen; Berj L Bardakjian Journal: J Neural Eng Date: 2015-03-13 Impact factor: 5.379