Epilepsy is a debilitating neurological condition that completely disrupts the lives of those who suffer from it and those close to them.
Over the years, technology has started to work hard to try to stem the effects of epilepsy attacks or to prevent them in some way.
Today there is a technological innovation that can offer even more precision in predicting epileptic attacks at any time of the day.
A system of artificial intelligence manages to anticipate the arrival of convulsions with an accuracy rate of 99,6%, and does so up to an hour before their occurrence.
Developed by Hisham Daoud e Magdy Bayoumi from the University of Louisiana at Lafayette, combines EEG (electroencephalogram) technology and a predictive model based on machine learning.
Early versions of this system used the two techniques in separate steps: today everything happens in parallel, making the forecasting process much less cumbersome.
By combining the analysis and classification processes into a single automated flow, Daoud and Bayoumi were able to obtain faster and more accurate predictions.
This system can be of great help to epileptic patients who in many cases can control seizures through medication.
Laboratory tests
The researchers tested their system on 22 patients at Boston Children's Hospital. To achieve this result, the system must be "calibrated" from time to time on the individual patient by carrying out an initial monitoring period. The suitable places? Those outside the clinics: home and work, where epileptic attacks can mostly occur. EEG electrodes already on the market can be safely used.
The team is now working on a custom chip to work out the algorithms. “We are dedicating ourselves to designing an efficient device that takes advantage of this algorithm, which has optimal size, consumption and speed for home use,” added Daoud.
The seizure system was described in a study published in the journal IEEE Transactions on Biomedical Circuits and Systems.
