Machine learning based approach for seizure recurrence prediction after first unprovoked seizure using retrospective clinical brain MRI and clinical information

Abstract:

In this talk, I will talk about our recent work, “Machine learning based approach for seizure recurrence prediction after a first unprovoked seizure”. Seizure recurrence is a significant concern for individuals after their first unprovoked seizure. Untreated, up to 50% of patients develop epilepsy within two years of the initial seizure. Despite the absence of currently used predictors, approximately 30% of first-seizure patients experience a second seizure. Predicting seizure recurrence is challenging, especially in patients with otherwise unremarkable clinical histories and normal MRI scans. Prophylactic antiseizure medications are also prescribed in individuals who would not have had recurrent seizures without treatment. To address this, we developed a machine learning framework using MRI and clinical data to predict the chance of seizure recurrence. The brain asymmetry-specific features were constructed using our proposed methods from the FreeSurfer parameters extracted from the MRI data, and the clinical features were generated by assigning binary, ternary, or positive integer values to the corresponding clinical information, forming the basis for building the machine learning models. Using a cohort of 169 subjects, including both recurrence (n=145) and non-recurrence (n=24) groups, our models exhibit promising classification accuracy and effectiveness in predicting seizure recurrence. The model also validated using a separate dataset consists of 45 patients. This novel methodology holds the potential to enhance patient outcomes by improving the accurate prediction of the risk of seizure recurrence.

Bio:

Soumen Ghosh is a postdoctoral research fellow at the Centre for Advanced Imaging at the University of Queensland, Australia. He completed his PhD in developing machine learning techniques for neuroimaging analysis for epilepsy research from The University of Queensland in 2024. He has a master’s in information technology from the University of Hyderabad, India. He has authored several articles and abstracts on machine learning, neuroimaging, and epilepsy. He presented his research at the American Epilepsy Society platform presentation in December 2023 and will present at OHBM, Seoul, South Korea, in June 2024. He received the Springer Student Award 2017 at the International Conference on Pattern Recognition and Machine Intelligence. His research interest is machine learning and neuroimage analysis.