University of Minnesota
University Relations

Minnesota Supercomputing Institute

Log out of MyMSI

Research Abstracts Online
January 2009 - March 2010

Main TOC ....... College TOC ....... Next Abstract

University of Minnesota Twin Cities
Institute of Technology
Department of Biomedical Engineering

PI: Theoden Netoff

Seizure Prediction Using Cost-sensitive Support Vector Machines

Approximately three million Americans suffer from epilepsy but no treatment currently exists. A device that could predict a seizure and notify the patient of the impending event or trigger an antiepileptic device would dramatically increase the quality of life for those patients. These researchers have developed a patient-specific classification algorithm to distinguish between preictal (prior to seizures) and interictal (between seizures) features extracted from EEG (electroencephalogram) recordings. They have demonstrated that the classifier based on cost-sensitive Support Vector Machines can distinguish preictal from interictal with high sensitivity and specificity, when applied to linear features, such as power spectrum in nine different frequency bands. It is computationally expensive work to build and test seizure prediction CSVM models for a patient via double cross-validation and it has been successfully accomplished using MSI resources. The researchers plan to expand their analysis to a very large database of EEG recordings being built by the Mayo Clinic, at nearly 15 Terabytes and growing, so they anticipate even higher use of the supercomputers.

Group Members

Dongyon Kang, Undergraduate Student
Yun Sang Park, Graduate Student
Chris Warren, Research Associate