Epileptic seizures can vary in severity from momentary loss of awareness to uncontrolled jerking and convulsions
Epilepsy, a neurological disorder that affects roughly 2% of the world's population, manifests itself in brief and intense episodes called seizures. During epileptic seizures patients can suffer from memory loss/momentary loss of awareness or suffer violent convulsions - uncontrollable jerking and movement.
During such violent convulsions, patients often harm themselves by biting down on their tongues, and can suffer head injuries due to uncontrolled falling or thrashing themselves against sharp or dangerous objects. Epileptic seizures can happen anywhere and at any time - at work, whilst outside or while driving - therefore, for those who suffer epileptic seizures, proper care and preventative treatment is crucial.
For diagnosis of the type of epileptic seizure and subsequent treatment approach, an Electroencephalography (EEG) scan is done that records the electrical activity of the brain and an Electrocardiography (ECG) that records the electrical activity of the heart. The resulting EEG and ECG dataset is often extremely large and requires painstakingly careful review to identify epileptic seizure activity.
"It is extremely exciting to collaborate with data scientists to determine whether advances in machine learning can be used to reach the holy grail of predicting when a person is going to have an epileptic seizure."
Prof Terence O’Brien — Director of Neurology, Alfred Health
The Epileptic Seizure Detection Research Project, which is an ongoing collaborative project, aims to use Machine Learning techniques to isolate and identify epileptic seizure episodes and to classify the type of epileptic seizure, from EEG and ECG data. This identification and classification of epileptic seizures will help medical researchers to further understand the signs and predictors that lead to and cause epileptic seizures. The features and predictors of epileptic seizures, found within EEG and ECG data, can be used to train Machine Learning algorithms to predict upcoming epileptic seizures. This early warning prediction can be used in a personal monitoring device, such as a heart rate tracker found in many of today's “smart watches”, that will help patients to prepare for and hopefully negate any harmful consequences of an epileptic seizure.