Scientists have developed a wearable device that detects epileptic seizures before they occur Thanks to the new solution, patients will have an hour more to prepare for the coming attack, which in some situations can even save their lives.

Researchers at Israel's Ben-Gurion University have developed a wearable device that will alert epileptic patients that they will experience an attack within an hour or so. This is a really breakthrough solution, because it gives a chance to prepare for a seizure attack, for example by going home or notifying a loved one, which in many cases may mean protection of health or even life. Especially that although some people actually respond to treatment that significantly reduces the frequency of seizures, 30% of patients suffer from drug-resistant epilepsy, i.e. epilepsy where, despite the use of drugs, seizures cannot be stopped.

This is where a device called Epiness comes into play, i.e. wearable that uses artificial intelligence and algorithms that analyze the electroencephalogram (EEG) in real time. If the software detects patterns suggesting an impending attack, patients will be informed about it via the smartphone application. The system itself consists of two main parts, a series of head-mounted EEG electrodes that monitor the brain's electrical activity, connected to a microprocessor responsible for machine learning algorithms. These have recently been trained on a large EEG database of epilepsy patients, covering both times without and during seizures.

By learning what patterns were present in the brain's electrical activity that preceded the attack, the algorithms were able to predict incoming seizures with an efficiency of about 97%. Algorithms are also able to techboss the brain's electrical signals from distracting background noise, which means that ultimately only a few electrodes are needed to maintain the effectiveness of the 'diagnosis'. - Epileptic seizures expose patients to many avoidable hazards, including falls, burns and other injuries. So we are very excited that the machine learning algorithms we have developed are able to effectively predict incoming attacks, even an hour before they happen. And since the algorithms have allowed for a significant reduction in the number of EEG electrodes, the device is both effective and user-friendly, says researcher Dr. Oren Shriki. One of the university companies, NeuroHelp, has a license for the Epiness device, thanks to which the product will go further and will be able to conduct clinical trials using the prototype later this year. We are waiting for some good news!