Tomsk Scientists Develop AI to Accelerate Epilepsy and Sleep Research

Researchers in Tomsk have created an AI-powered system that can decode brain activity in seconds. The breakthrough could transform the study and treatment of epilepsy and sleep disorders.

Scientists at Tomsk State University, working with the Institute of Higher Nervous Activity and Neurophysiology of the Russian Academy of Sciences, have unveiled a pioneering development in digital medicine: a system for the automated analysis of brain pathologies using artificial intelligence.

The team is currently conducting preclinical studies on absence epilepsy and sleep disorders in laboratory animals. In the future, the technology could revolutionize the diagnosis and therapy of serious neurological diseases in humans.

The system tackles one of the most labor-intensive tasks in neurophysiology: decoding electrocorticogram data. Traditionally, only highly trained specialists can handle this work manually, requiring hours of effort. The algorithm, however, can identify deep sleep phases and detect brief episodes of absence epilepsy with extreme precision. These episodes are notoriously difficult to diagnose, as they involve no seizures and appear only as short lapses in consciousness.

The AI model was trained on an exclusive dataset annotated by leading neurophysiologists. The application performs a full cycle of electrocorticogram analysis automatically: one hour of recordings can be decoded in just 25 seconds. This opens new possibilities for understanding and treating one of the world’s most common neurological disorders, which affects around 50 million people, according to the WHO.