Project 458964

Epilepsy is the most common serious neurological disorder of childhood, characterized by a propensity for recurrent seizures. Children with epilepsy have higher mortality than the general population, frequent hospital visits, difficulties in school and suffer from social isolation. For some children, a surgical treatment, vagus nerve stimulation (VNS), may reduce their seizures and significantly improve quality of life. VNS is a device that is implanted in the neck to deliver electrical stimulation that results in reduced seizure frequency. Only 25-50% of children achieve meaningful benefit following implantation. Despite over 100,000 procedures performed worldwide, there are no means to identify those who will benefit from surgical treatment before surgery. Therefore, a large proportion of children undergo the risks of surgery unnecessarily. Furthermore, the inability to identify which child is most likely to benefit from surgery results in poor resource allocation within the Canadian healthcare system. There is, therefore, an urgent unmet need to develop novel means to pre-surgically predict outcomes following VNS. I will leverage advances in machine learning and brain network mapping to study brain differences in children who do, and do not, respond to VNS. Through advanced computational processing, I will develop the first predictive algorithm to identify children who are most likely to benefit from VNS. I will first collect data from high-volume epilepsy centres across North America. Next, advanced machine learning models will be trained to develop a novel prediction algorithm. Subsequently, this will be applied to prospective datasets to validate these findings and translate these discoveries to clinical care. By applying machine learning models to the study and analysis of brain network circuitry, my research will inform optimal candidate selection for VNS, thereby reducing unnecessary surgery and improving precious Canadian healthcare resource allocation.