Prediction of epilepsy seizures from neurological crises

Epilepsy disease, known in the community as ‘sara’, is a serious and widespread neurological disease affecting 0.4-1% of the world population according to the World Health Organization (WHO) estimates. Epilepsy disease, characterized by instantaneous and repetitive seizures, affects people in almost every age group, although it occurs more frequently in childhood and adulthood. In general, loss of consciousness, movement disorder, such as only seizures and seizures that affect a period of several hours following the seizure, but can be controlled by drugs creates temporary conditions. It is also difficult to diagnose non-epileptic seizures (pseudo-or pseudo-seizures) that are similar to epileptic seizures. Whether the seizures of epileptic patients are epileptic (frequent and reliable diagnostic method is video-EEG measurement at the time of crisis) and the dose of the drugs to be used are determined based on the patient's history. It was determined that 10-20% of the patients who applied to the specialist physician with the suspicion of having epileptic seizure did not have. Based on the patient's response to epileptic drugs, its detection can be determined on average 7.2 years after the start of treatment. In this thesis, an integrated approach with the developing technology is proposed for the prediction of epileptic seizures with the help of various sensors (such as EMG, ECG, Accelerometer). The main purpose of the study is to create a test routine to reduce the cost of the procedure for the prediction of epileptic seizures. Using a data set taken from patients treated in the Dicle University, Faculty of Medicine, Department of Neurology, during an epileptic seizure process, it was optimized by various methods and classified by over-learning machines (ELM) with later characteristics.