Artificial intelligence predicts epileptic seizures with almost complete accuracy

Nearly 50 million people worldwide have epileptic seizures, often unannounced. Two researchers from the University of Louisiana, however, have developed a new model of artificial intelligence that can predict crises up to an hour before they even occur with 99.6% accuracy!

Because of the unexpected and sudden onset, epileptic seizures have a heavy psychological burden on patients, especially at the social level. Their early detection and warning to the patient can significantly improve their quality of life, giving them enough time to take action. Seizures can be controlled with medication for up to 70% of patients.

The new artificial intelligence system is fully automated, unlike previous methods where it required manual data extraction and categorization. With the deep learning algorithm, the system can be adapted to the brain activity of each patient, increasing its accuracy. Applying the system to the patient requires only a few hours of non-invasive testing, as long as the artificial intelligence model is trained and can be performed outside hospital facilities through commercially available wearable electrocardiograms.