Machine learning-based characterisation of high-frequency oscillations as biomarkers of post-traumatic epilepsy

Post-traumatic epilepsy (PTE) is a serious and unpredictable complication of traumatic brain injury (TBI). Identifying reliable electrophysiological biomarkers that predict epilepsy development remains a major unmet need. High-frequency oscillations (HFOs; 80-500 Hz) have emerged as promising candidate biomarkers; however, their detection remains technically demanding and prone to subjectivity, as accurate identification often requires manual validation and is limited by inter-rater variability. This Honours project will leverage pre-existing long-term EEG recordings from a PTE rat model. The dataset includes animals assigned to sham surgery, TBI without epilepsy development, and TBI with subsequent epilepsy. The student will focus on computational and machine learning approaches to improve HFO detection and characterisation. Specifically, the project aims to: 1. Develop and evaluate machine learning classifiers to distinguish true HFOs from false positives. 2. Determine whether machine learning-derived HFO metrics differentiate experimental groups (sham, TBI- epilepsy, TBI+ epilepsy), thereby assessing their potential as biomarkers of epileptogenesis. The student will gain training in EEG signal processing, feature extraction, supervised machine learning, and statistical analysis. This project provides an opportunity to contribute to translational biomarker development in epilepsy research while working with a high-quality preclinical dataset.