Eriksson, Maria Helena; (2023) Clinical and cognitive outcomes after neurosurgical treatment for epilepsy in children. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text thesis_Eriksson_final.pdf - Other Download (17MB) | Preview

Abstract

Epilepsy surgery is a low-risk and effective form of treatment for carefully selected children with drug-resistant epilepsy. Still, there is concern that it remains underutilised, and that seizure freedom rates have not improved over time. It also remains a considerable challenge for clinicians to predict which patients will be rendered seizure- free through surgery, and which will not, before the procedure has been carried out. Furthermore, although the primary goal of epilepsy surgery is to provide seizure freedom, there exists a hope that it will also improve the cognitive functioning of children with drug-resistant epilepsy; however, findings to date have been variable. This thesis had three main aims, namely to (1) take stock of what has changed in paediatric epilepsy surgery over the past two decades, (2) apply machine learning technology to the prediction of seizure outcome after epilepsy surgery, and (3) map the long-term neuropsychological trajectories of children undergoing epilepsy surgery, to determine if surgery has the potential to improve these outcomes. To address these aims, I built a large, retrospective cohort of children who had undergone epilepsy surgery at Great Ormond Street Hospital for Children. I analysed trends in surgical practices, patient characteristics, and post-operative outcomes using regression analyses. I then trained three models – a logistic regression and two machine learning models – to predict seizure outcome. I also evaluated the performance of a recently published machine learning model on the same patient cohort. Lastly, I analysed long-term pre- and post-operative neuropsychological functioning – measured up to 12 years before and 16 years after surgery. I found that, despite increases in referral and surgical volumes, children with drug- resistant epilepsy continue to be put forward for surgery late. Seizure freedom rates have, however, improved, but only if the concurrent increase in complex cases is accounted for. The three models trained to predict seizure outcome performed similarly well, and significantly better than the external model, suggesting that machine learning technology alone will not improve our ability to predict outcome. Finally, children showed significant declines in neuropsychological functioning in the time leading up to surgery. However, by providing seizure freedom, surgery was able to reverse this downward trajectory.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item