Mark J Cook1, Philippa J Karoly1,2, Dean R Freestone1, David Himes3, Kent Leyde3, Samuel Berkovic4, Terence O'Brien5, David B Grayden1,2, Ray Boston1. 1. Departments of Medicine, St. Vincent's Hospital, University of Melbourne, Fitzroy, Victoria, Australia. 2. Department of Electrical and Electronic Engineering, University of Melbourne, Parkville, Victoria, Australia. 3. NeuroVista Corporation, Seattle, Washington, U.S.A. 4. Austin and Repatriation Medical Centre, Heidelberg, Victoria, Australia. 5. Royal Melbourne Hospital, Parkville, Victoria, Australia.
Abstract
OBJECTIVE: We report on a quantitative analysis of data from a study that acquired continuous long-term ambulatory human electroencephalography (EEG) data over extended periods. The objectives were to examine the seizure duration and interseizure interval (ISI), their relationship to each other, and the effect of these features on the clinical manifestation of events. METHODS: Chronic ambulatory intracranial EEG data acquired for the purpose of seizure prediction were analyzed and annotated. A detection algorithm identified potential seizure activity, which was manually confirmed. Events were classified as clinically corroborated, electroencephalographically identical but not clinically corroborated, or subclinical. K-means cluster analysis supplemented by finite mixture modeling was used to locate groupings of seizure duration and ISI. RESULTS: Quantitative analyses confirmed well-resolved groups of seizure duration and ISIs, which were either mono-modal or multimodal, and highly subject specific. Subjects with a single population of seizures were linked to improved seizure prediction outcomes. There was a complex relationship between clinically manifest seizures, seizure duration, and interval. SIGNIFICANCE: These data represent the first opportunity to reliably investigate the statistics of seizure occurrence in a realistic, long-term setting. The presence of distinct duration groups implies that the evolution of seizures follows a predetermined course. Patterns of seizure activity showed considerable variation between individuals, but were highly predictable within individuals. This finding indicates seizure dynamics are characterized by subject-specific time scales; therefore, temporal distributions of seizures should also be interpreted on an individual level. Identification of duration and interval subgroups may provide a new avenue for improving seizure prediction. Wiley Periodicals, Inc.
OBJECTIVE: We report on a quantitative analysis of data from a study that acquired continuous long-term ambulatory human electroencephalography (EEG) data over extended periods. The objectives were to examine the seizure duration and interseizure interval (ISI), their relationship to each other, and the effect of these features on the clinical manifestation of events. METHODS: Chronic ambulatory intracranial EEG data acquired for the purpose of seizure prediction were analyzed and annotated. A detection algorithm identified potential seizure activity, which was manually confirmed. Events were classified as clinically corroborated, electroencephalographically identical but not clinically corroborated, or subclinical. K-means cluster analysis supplemented by finite mixture modeling was used to locate groupings of seizure duration and ISI. RESULTS: Quantitative analyses confirmed well-resolved groups of seizure duration and ISIs, which were either mono-modal or multimodal, and highly subject specific. Subjects with a single population of seizures were linked to improved seizure prediction outcomes. There was a complex relationship between clinically manifest seizures, seizure duration, and interval. SIGNIFICANCE: These data represent the first opportunity to reliably investigate the statistics of seizure occurrence in a realistic, long-term setting. The presence of distinct duration groups implies that the evolution of seizures follows a predetermined course. Patterns of seizure activity showed considerable variation between individuals, but were highly predictable within individuals. This finding indicates seizure dynamics are characterized by subject-specific time scales; therefore, temporal distributions of seizures should also be interpreted on an individual level. Identification of duration and interval subgroups may provide a new avenue for improving seizure prediction. Wiley Periodicals, Inc.
Authors: Philippa J Karoly; Ewan S Nurse; Dean R Freestone; Hoameng Ung; Mark J Cook; Ray Boston Journal: Epilepsia Date: 2017-01-13 Impact factor: 5.864
Authors: Gabrielle M Schroeder; Fahmida A Chowdhury; Mark J Cook; Beate Diehl; John S Duncan; Philippa J Karoly; Peter N Taylor; Yujiang Wang Journal: Brain Commun Date: 2022-07-06
Authors: Daniel E Payne; Katrina L Dell; Phillipa J Karoly; Vaclav Kremen; Vaclav Gerla; Levin Kuhlmann; Gregory A Worrell; Mark J Cook; David B Grayden; Dean R Freestone Journal: Epilepsia Date: 2020-12-30 Impact factor: 6.740
Authors: Maxime O Baud; Jonathan K Kleen; Emily A Mirro; Jason C Andrechak; David King-Stephens; Edward F Chang; Vikram R Rao Journal: Nat Commun Date: 2018-01-08 Impact factor: 14.919
Authors: Daniel M Goldenholz; Joseph J Tharayil; Rubin Kuzniecky; Philippa Karoly; William H Theodore; Mark J Cook Journal: Epilepsia Open Date: 2017-01-18