A Hierarchical Failure Model for Characterizing Progressive Abnormal Neuronal Network Interactions During Epileptic Seizure Evolution

• Activity Number: 77
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Catherine Stamoulis*+ and Bernard S Chang
• Companies: Harvard Medical School and Harvard Medical School
• Keywords: Hierarchical failure models ; High-dimensional data ; Networks ; Information theory

Abstract

Epilepsy is a common neurological disorder that affects more than 50 million people worldwide. The hallmark of the disorder is seizures, transient events of abnormal neuronal network coordination. Seizure evolution may be thought of as a hierarchical failure process, where aberrant network synchronization corresponds to 'failure' of individual system components, and seizure onset corresponds to the time of combinatorial network failure. A hierarchical time-to-failure model has been developed, to describe progressive abnormal network coordination leading to seizures. The brain was assumed to be a multi-component dynamic system, represented by a weighted directional network, in which individual edge weights correspond to baseline neural connections independent of seizures. Connectivity was quantified by information parameters estimated from electroencephalographic (EEG) data. Failure of individual components was defined as the time(s) at which edge weights between adjacent nodes exceeded brain region-specific thresholds estimated from baseline EEGs. Times-to-failure were estimated from pre-seizure EEGs and were found to follow a power-law distribution prior to seizure onset.