Researchers Develop AI to Detect Hidden Signs of Epilepsy
A new machine learning algorithm is proving capable of identifying subtle patterns of epilepsy in brain activity, even when a patient is not experiencing a seizure.
The algorithm identified genetic markers of epilepsy with high accuracy using only "normal" brain activity segments.
The Breakthrough
Developed by researchers at the University of Delaware and Nemours Children's Health, the algorithm analyzes EEG data to detect neurological abnormalities that are otherwise invisible to the human eye.
The study focused on mice with the TSC1 gene variation, a known cause of epilepsy. In a proof-of-concept, the algorithm successfully distinguished between different mouse strains and detected the epilepsy-linked gene with high accuracy in two of three test groups. Critically, this was achieved using EEG segments that contained no seizure activity.
The findings were published in the Journal of Neural Engineering.
The Next Phase: From Mice to Children
The research team is now moving to human trials.
- Funding: The next stage is supported by the Delaware Clinical and Translational Research ACCEL Program.
- Focus: The team will test the algorithm on EEGs collected from children undergoing epilepsy evaluation at Nemours Children's Health.
- Goal: To identify reliable biomarkers—early warning signs—that indicate underlying changes in brain activity before seizures occur, paving the way for earlier detection and intervention.
Broader Implications for Neurology
The potential for this AI-driven approach extends well beyond epilepsy.
Researchers suggest similar algorithms could be adapted to detect patterns associated with autism and ADHD.
Looking ahead, this technology could be integrated into wearable EEG devices, allowing for continuous, in-home monitoring. This would enable a new era of precision medicine, where treatment decisions are guided by real-time, subtle shifts in brain activity rather than relying solely on observable symptoms.