Reliable Seizure Prediction Using Physiological Signals and Machine Learning

点击翻译按钮获取中文摘要

For most individuals living with epilepsy, seizures are relatively infrequent events occupying a small fraction of their life. Despite spending as little as 0.01% of their lives having seizures (typically only minutes per month), people with epilepsy take anti-seizure drugs (ASD) daily, suffer ASD related side effects, and spend their lives dreading when the next seizure will strike. The apparent randomness of seizures is associated with significant psychological consequences. In addition, despite daily ASD, approximately 1/3 of patients continue to have seizures. We hypothesize that epilepsy can be more effectively treated, both the seizures and their psychological impact, by providing patients with real-time seizure forecasting. There is strong evidence that focal epilepsy is associated with a variable seizure risk that may enable adaptive therapy targeting periods of high seizure probability. Periods of low seizure probability could require lower ASD doses, reducing exposure and side effects. We propose that high seizure probability states will respond to adaptive electrical brain stimulation (aEBS). In addition, patients could alter their activities during periods of high seizure probability to reduce injury and manage their ASD and activities. The hypotheses driving this proposal are that 1.) seizures can be prevented (reduced incidence) by targeted EBS therapy during the pre-ictal state 2.) seizures are not random events, and that brain states associated with low and high seizure probability can be reliably classified using machine learning methods applied to physiologic signals and used to adaptively change EBS parameters. 3.) Furthermore, we propose forecasting can be improved using multi-modal features beyond passive iEEG recordings, including active brain probing with electrical stimulation (impedance & evoked potentials), core temperature, ECG and serum immunological markers. Goal: Develop reliable seizure forecasting (>90% sensitivity) with few false positives (<1% time in warning) and demonstrate modulation of seizure risk and reduction of focal seizures using aEBS.