"The algorithm successfully distinguished between different mouse strains... identifying a variation in the TSC1 gene with high accuracy."

Spotting Seizures Before They Start

Researchers from the University of Delaware (UD) and Nemours Children's Health have developed a machine learning algorithm capable of detecting subtle patterns in electroencephalogram (EEG) recordings associated with epilepsy—even in the absence of visible seizure activity.

Methodology and Findings

In a proof-of-concept study conducted on mice, the algorithm analyzed baseline EEG segments that did not contain seizure activity.

Published in the Journal of Neural Engineering, the study reports that the algorithm successfully distinguished between different mouse strains. Furthermore, it identified a variation in the TSC1 gene (which is linked to epilepsy) with high accuracy in two out of three mouse strains tested.

Future Research Plans

The research team plans to test the algorithm on EEG recordings obtained from children undergoing epilepsy evaluations at Nemours Children's Health. This next phase of research is funded by the Delaware Clinical and Translational Research ACCEL Program.

The stated goal is to identify biomarkers that indicate underlying changes in brain activity before seizures occur. Researchers aim to enable earlier detection and treatment of epilepsy.

Broader Context and Potential Applications

The researchers have suggested that similar machine learning approaches could be applied to other neurological conditions, including autism and attention-deficit/hyperactivity disorder (ADHD).

Potential future applications include:

• Continuous monitoring via wearable EEG devices
• Using algorithms to inform precision medicine decisions regarding treatment