Sign In

Communications of the ACM

ACM TechNews

Study Looks at Better Prediction For Epileptic Seizures Through Adaptive Learning Approach


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
University of Texas at Arlington professor Shouyi Wang.

A new computational model uses a patient's personalized medical information to predict when an epileptic seizure will occur.

Credit: University of Texas at Austin

University of Texas (UT) at Arlington researchers have developed a computational model they say can more accurately predict when an epileptic seizure will occur next based on the patient's personalized medical information. The model analyzes electroencephalography (EEG) readings from an individual to predict future seizures.

UT Arlington professor Shouyi Wang says that such an early warning system could lead a patient to use medicine to combat an oncoming seizure.

Early testing indicates the computational model could provide 70-percent accuracy or better and give a prediction horizon of about 30 minutes before the actual seizure would occur. The current model collects data through a cap embedded with EEG wires, but the researchers are developing a less-obtrusive EEG cap that will record and transmit readings to a box for data download or transmission.

"This computational model might be used to predict other life-threatening episodes of diseases," says UT Arlington professor Victoria Chen.

Wang says the model builds upon an adaptive-learning framework and is capable of achieving increasingly accurate prediction performance for each individual patient by collecting more personalized medical data.

From UT Arlington News Center
View Full Article

 

Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA


 

No entries found