A new AI-powered model can predict the occurrence of epileptic seizures up to an hour before onset with 99.6% accuracy, according to US-headquartered technical professional organization IEEE.
Developed by researchers Hisham Daoud and Magdy Bayoumi of the University of Louisiana, the system was tested on 22 patients at the Boston Children’s Hospital. Daoud says that to continually achieve high accuracy, there will be a need to train algorithms on individual patients in the future, rather than pursuing a “one-size fits all” model,
With each patient having unique brain patterns, as more are tested, there will be increased complexity in the prediction models. The small sample size from the research is exciting in the medical field. Going forward, the training could take a few hours of non-invasive EEG. It will monitor a patient around times that they are having a seizure to ensure it adapts to their individual patterns.
The software component is now complete. Coming up next is creating customized chips that can process the algorithms. The researchers need more efficient hardware allowing for scalable system size and power consumption. This will help them to create a practical application comfortable for patients.
Around 60 million worldwide have epilepsy. Seizures are known to happen suddenly, without warning and vary by person. The novel AI model could greatly improve the quality of patients’ lives, giving them enough time to take action.
