An AI system based on a machine learning algorithm for analyzing EEG signals
Epileptic seizures expose everything about people’s lives from various preventable hazards, including falls, burns, and other injuries to restricting independence. One of the debilitating things about epilepsy is the uncertainty of knowing when a seizure is going to occur. Unfortunately, current seizure-predicting devices that can alert patients are still absent. These devices can detect a seizure in real-time but fail to provide advanced warnings of impending seizures.
However, Epiness, a first of its kind device for detecting and predicting epileptic seizures has been developed. A device that can generate an advanced warning about an upcoming seizure and sends to a smartphone up to an hour prior to its onset. Now, patients can be prepared for upcoming seizures.
Developed by researchers at Ben-Gurion University of the Negev (BGU), Epiness functions based on a new, ground-breaking combination of EFG-based monitoring of brain activity together with proprietary machine-learning algorithms. The device combines a wearable EEG device with software that minimizes the number of necessary EEG electrodes and optimizes electrode placement on the scalp.
Epilepsy is a neurological disorder that affects a patient’s way of life. The seizures occur when there is a sudden abnormal electrical activity that temporarily interrupts normal brain function. Up to 30% of patients live under constant fear of impending seizures and ask if they will be able to predict seizures, says IBM Research.
Predicting impending seizures
Epiness offers a substantial improvement in the quality of life, enabling patients to avoid seizure-related injuries. The machine-learning algorithms are designed to filter noise that is not related to brain activity, extract informative measures of the underlying brain dynamics, and distinguish between brain activity before an expected epileptic seizure and brain activity when a seizure is not expected to occur.
Researchers at BGU developed and tested this new algorithm using EEG data from a large dataset of people with epilepsy that were monitored for several days prior to surgery. The patients’ recorded data were divided into short segments that were either preictal (pre-seizure) or inter-ictal. Several machine learning algorithms with differing complexities were trained on pre-allocated training data (comprising 80% of the initial EEG data), and their prediction performance, as well as electrode-dependent performance, was assessed on the remaining 20% of the data. The algorithm with the best prediction performance reached a 97% level of accuracy, with near-optimal performance maintained (95%) even with relatively few electrodes.