Multidimensional CNN and LSTM for Predicting Epilepsy Seizure Activities

Abstract

Epilepsy is a chronic neurological disease caused by sudden abnormal brain discharges, leading to temporary brain dysfunction. It can manifest in various ways, including paroxysmal movement, sensory, autonomic nerve, awareness, and mental abnormalities. It is now the second largest neurological disorder worldwide, affecting around 70 million people and increasing by approximately 2 million new cases each year. While about 70% of epilepsy patients can control their seizures with regular antiepileptic drugs, surgery, or nerve stimulation treatments, the remaining 30% suffer from intractable epilepsy without effective treatment, causing significant burden and potential danger to their lives. Early prediction and treatment are crucial to prevent harm to patients. Electroencephalogram (EEG) is a valuable tool for diagnosing epilepsy as it records the brain's electrical activity. EEG can be divided into scalp and intracranial types, and doctors typically analyze EEG signals of epileptic patients into four periods.