Patient-Specific Electrical Stimulation to Effectively Suppress Seizures using a Data-Driven Dynamical Network Model

Seizure suppression is critical for enhancing the quality of life for patients with epilepsy. However, nearly 15 million patients have medically refractory epilepsy, meaning their seizures cannot be suppressed with medication. Electrical stimulation therapy is an alternative therapy to suppress seizures. However, determining where and how to stimulate remains an open problem. Currently, physicians rely on trial-and-error, which is inefficient. In this work, we present a data-driven patient-specific method that identifies the best location to stimulate and designs the stimulation signal to suppress seizures by leveraging the patient's cortico-cortical evoked potentials (CCEPs). Specifically, we construct transfer function models from the CCEPs recordings and plot the magnitude response versus frequency. From this bode plot, we obtain the frequency that minimizes the magnitude response in the seizure onset zone (SOZ) channels. In silico, we demonstrate that our method suppresses signals in the clinically annotated SOZ by up to 81%. Our method lays the groundwork for neurostimulation to be an effective treatment for medically refractory epilepsy. Future work will focus on prospectively validating this method in the clinic.