Causal brain activity modeling for the understanding of epilepsy pathogenesis

Abstract

Epilepsy affects 1% of the world’s population yearly and is one of the most widespread diseases. Although some forms of epilepsy are effectively treated using a medication, nearly every third patient needs surgical intervention to remove the epileptogenic area. This study examined an epileptic brain’s activity during the hallmarks of epileptic activity - interictal epileptiform discharges (IEDs) - using Causal Bayesian Networks. Results showed that trends of causal activity increase before an IED and decrease afterwards, differentiating across awareness states. Another benefit of the study is the recommendation system for a clinician while evaluating the epileptogenic region. The channels that have most frequently been a cause of an IED are reported as a potential resection area. The code is freely available on the Github repository with the corresponding documentation for a clinician.