Wednesday 09 April 2025
A new approach to detecting seizures in people with epilepsy has been developed, using a type of artificial intelligence that can learn and adapt on its own.
The researchers behind the work used a technique called neural additive models, which combines multiple machine learning algorithms to improve performance. The model is trained on data from patients who have had seizures, allowing it to learn patterns and characteristics associated with seizure activity.
One of the key benefits of this approach is that it can be adapted for use in real-time, making it potentially useful for wearable devices or implantable sensors that could detect seizures as they occur. This could allow for faster and more effective treatment, reducing the risk of injury or death.
The model was tested on data from 24 patients with epilepsy, and was able to accurately detect seizures in all but one case. It also performed well when tested on new, unseen data – a key test of its ability to generalize to real-world scenarios.
Another advantage of this approach is that it can be used to detect seizures in people who have not had them before, allowing for earlier diagnosis and treatment. This could potentially improve outcomes for people with epilepsy, who currently often face a long and uncertain journey to diagnosis.
The model works by using a combination of features extracted from the brain activity data, including changes in frequency and amplitude of different brain wave patterns. It then uses these features to make predictions about whether a seizure is likely to occur or has already occurred.
The researchers believe that this approach could have far-reaching implications for the treatment of epilepsy, allowing for more accurate and timely detection and potentially reducing the risk of seizures occurring in the first place. They are now working on refining the model and testing it in real-world scenarios.
The development of this technology is a significant step forward in the quest to better understand and treat epilepsy, and could have a major impact on the lives of people living with the condition.
Cite this article: “Unlocking Real-Time Seizure Detection: A Novel Approach Combining Neural Additive Models and Online Updating Techniques”, The Science Archive, 2025.
Epilepsy, Artificial Intelligence, Machine Learning, Neural Additive Models, Seizure Detection, Brain Activity, Wearable Devices, Implantable Sensors, Diagnosis, Treatment.
