170 Early Electroencephalography Biomarkers of Cortical Dysfunction to Predict Long-Term Risk of Post-Traumatic Epilepsy

INTRODUCTION: Up to one-third of severe traumatic brain injury (TBI) patients develop post-traumatic epilepsy (PTE) and early electroencephalography (EEG) biomarkers may allow for identification of PTE risk, guiding anti-epileptogenic therapies. METHODS: We retrospectively analyzed a prospective database of severe TBI patients treated at a single level one trauma center from 2012 through 2018. We identified a cohort of patients who survived to two years and were outcome matched using age and the Glasgow Outcomes Scale Extended (GOSE). We used continuous EEG collected within the first five days post-trauma. We developed a novel set of EEG features to quantify focal dysfunction through computing MAD of the PSD of the canonical EEG frequency bands across channels (delta, theta, alpha, beta). The MAD quantifies how variable each band is across the electrodes by dividing by the variability in all bands. In focal dysfunction, bursts of delta or theta in a group of electrodes increase the variability of delta or theta, increasing delta MAD. We then developed a support vector machine to predict long-term PTE risk using delta and theta MAD and average spectral power. RESULTS: We identified 21 patients with PTE and 20 without who survived two years post-injury. The median time to onset of PTE was 7.2 months post-trauma and GOSE was similar when stratified by PTE at 6- and 12-months (p>0.73). Validation accuracy was 84%, sensitivity 70%, specificity 86%, and area under the receiving operating curve of 0.80. CONCLUSIONS: We developed a novel set of early EEG biomarkers to measure focal cortical dysfunction that accurately predicted long-term PTE risk early after trauma.