EAN 2026 | The development of a deep learning tool to detect drug-resistant epilepsy with EEG
Manuel Toledo, MD, PhD, VHIR Vall d’Hebron Institut de Recerca, Barcelona, Spain, shares insights into the development of a deep learning tool to detect drug-resistant epilepsy with EEG. Dr Toledo explains how this approach has the potential to accelerate diagnosis, improve the identification of drug resistance, and support more informed treatment decisions. This interview took place at the 12th Congress of the European Academy of Neurology (EAN) in Geneva, Switzerland.
These works are owned by Magdalen Medical Publishing (MMP) and are protected by copyright laws and treaties around the world. All rights are reserved.
Transcript
We are working with the EEG for the last 100 years, one century, and it’s all based on personal experience of physicians able to read the EEG. However, the EEGs are only images, images that represent the brain activity. And we know with that tool, we are able to make a diagnosis of epilepsy or even have the perspective how the brain is working then. So I guess that using massive tools to read EEGs like deep learning and the AI methods, that will be of real interest to be able to diagnose patients with epilepsy right away and overall to decide whether they will be resistant or they will be responsive to different treatments...
We are working with the EEG for the last 100 years, one century, and it’s all based on personal experience of physicians able to read the EEG. However, the EEGs are only images, images that represent the brain activity. And we know with that tool, we are able to make a diagnosis of epilepsy or even have the perspective how the brain is working then. So I guess that using massive tools to read EEGs like deep learning and the AI methods, that will be of real interest to be able to diagnose patients with epilepsy right away and overall to decide whether they will be resistant or they will be responsive to different treatments. So basically, I’m talking to a biomarker which is patent, and we have been using that, and we have all the data collected to decide whether the patient has epilepsy or not. That will be the short future coming, and I think that will be a big difference instead of waiting on an EEG on the interpretation of individual doctors to be able to say whether they have epilepsy or not.
Nowadays, the definition of drug-resistant epilepsy is a little bit, let’s say, like categoric. It’s like whether you have drug-resistant epilepsy or not, and then there are different degrees when we talk among scientific or even when we talk to someone else. We say this person has a severe epilepsy or a mild epilepsy. It doesn’t exist scientifically. And then we consider whether those patients had drug-resistant or not depending on whether they have recurrent seizures or not. But then there are other aspects of the disease, like the severity, like generalized tonic-clonic seizures, or if they have cognitive impairment, or they have psychiatric or behavioral disorders, that are not considered within the definition of drug-resistant epilepsy. We should scale that definition into different degrees of severity. And from there, we should be able to decide with different biomarkers where those patients will be, like more or less responsive, or even to take actions to be more or less aggressive on the treatments from the very beginning, instead of waiting for one or two years to decide whether that patient should have been treated as more aggressive or not.
This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.
Read more...
