Vagus nerve stimulation therapy outcome prediction using brain connectivity estimators

Epilepsy is a neurological disorder that affects millions of people worldwide. The most common treatments for epilepsy are the use of antiepileptic drugs (AEDs) and surgery in more severe cases, but these treatments may not be sufficient to treat some patients and then vagus nerve stimulation (VNS) therapy can be used as an add-on treatment to reduce seizures. However, a question that arises with the use of this therapy is the prediction of patient response to treatment, which would avoid unnecessary implantation of VNS in non-responders. In this master's thesis we used functional connectivity measures in combination with graph measures and machine learning models to predict the response of patients to the therapy. Although the study of the influence of the VNS on functional connectivity is not new, this work is distinguished by using only electroencephalograms recorded before VNS implantation and a wider range of functional connectivity and graph measures. After calculating the measures in a population of 37 patients, we found significant differences (p<0.05) between the global efficiency, average clustering coefficient, and modularity of responders and non-responders, indicating that these measures might work as biomarkers to predict the response to the therapy. Moreover, using a subset of the measures as features we were able to train a machine learning model that achieved a validation accuracy of 0.87 and a test accuracy of 0.86.