AI Model Introduced for Transparent Epilepsy Diagnosis
The system explains how it reaches conclusions, addressing a key barrier to AI adoption in clinical settings.
Researchers from Plekhanov Russian University of Economics and Pirogov National Medical and Surgical Center have developed a new AI-based method for diagnosing psychoneurological disorders. The neural network not only analyzes electroencephalograms and identifies a patient’s condition but also provides detailed explanations of the reasoning behind its conclusions, offering additional support for patients with epilepsy.
The approach addresses one of the main obstacles to broader AI adoption in medicine. Clinicians are often reluctant to rely on machine-generated diagnoses without understanding how they are produced. The new system makes the diagnostic process fully transparent.
The method was tested on three different neural network architectures, including a hybrid model that delivered the strongest performance. Feature analysis confirmed alignment with established neurophysiological mechanisms of epilepsy.
