An AI system for forecasting of an epileptic seizure

2019: Announcement

In the middle of November, 2019 the system of artificial intelligence which can predict epileptic seizures with an accuracy of 99.6% was provided. Moreover, it is capable to predict their development for an hour before emergence of the main symptoms. Such exact forecast allows people to be prepared for the attack in time and to take medicine. The solution already began to be used in practice.

According to WHO data, by November, 2019 about 50 million people around the world have epilepsy, and 70% of these patients can control attacks using the taken medicine in time. Thus, the high-precision forecast can significantly improve quality of life of patients.

The system of artificial intelligence which can predict epileptic seizures with an accuracy of 99.6% was provided

The new system of artificial intelligence developed by Hisham Daoud and Magdy Bayoumi from the University of Louisiana in Lafayette took a serious step forward in comparison with sshchestvuyushchy forecasting methods. Techniques now in use allow to analyze activity of a brain using EEG, and then to estimate using predictive model.

However at each person patterns of excitation of a brain are unique that complicates exact forecasting of spasms. The previous models were developed for carrying out two-stage process where patterns of a brain had to be taken manually that seriously complicated model. The new method performs both of these functions along with the help of an algorithm of deep learning which estimates activity of a brain and probability of distribution of excitation on pathological ways.

By November, 2019 the method is used in limited scope. The command works on the special chip which can help with mass processing of necessary algorithms. Researchers consider that the similar device will be extremely demanded by patients with epilepsy.^[1]

Notes