Epilepsy is a chronic neurological condition where abnormal electrical activity in the brain causes seizures. Epileptic seizures are events that can vary from brief and nearly unnoticeable periods to long stretches of vigorous shaking, rhythmic muscle contractions or muscle spasms. Approximately 50 million people worldwide suffer from epilepsy, making it one of the most common neurological diseases globally. The seizure causes an array of problems, especially for patients who have developed drug-resistant epilepsy, the possibility of predicting seizures in advance could be very useful, not only for the patients but also for the medical professionals dealing with such disorders.
In current practice trained neurologists analyze signals from EEG and synchronized video of the patients which tends to be a very tedious and slow task as it requires differentiating signals across multiple days. Here an automated system that accurately analyses patterns and classifies signal segments into different temporal dynamics of the brain, would be extremely fast and useful. Due to extensive development in the field of EEG data collection and machine learning, there is a potential to completely automate the seizure prediction process, in this endeavor we employ numerous strategies to clean the time-series data, perform feature engineering and apply various machine learning techniques to predict the onset of a seizure episode.