Time-Series Prediction of Epileptic Seizures Using EEG

Epileptic seizures occur when brain nerve cells exhibit abnormal electrical activity, sometimes causing loss of consciousness and convulsions. While epileptic seizures can be evaluated through EEG, they require long-term visual observation by highly trained specialists. Therefore, in emergency situations where specialists cannot respond, seizure detection may be delayed, leading to severe consequences. In this research, we are working on developing algorithms that can sequentially predict seizure probability by applying research on stochastic generative models of EEG. The physiological states within the brain differ before, after, and during epileptic seizures in patients. Therefore, we aim to accurately detect epileptic seizures by modeling the probability distributions of EEG output by each state and the transition processes between states. Technical approach: - Machine learning using features obtained from stochastic generative models - Extraction of patient-invariant features through domain adversarial learning We are advancing verification using real clinical data through joint research with Okayama University Hospital.