Predicting stroke severity with a 3-min recording from the Muse portable EEG system for rapid diagnosis of stroke
Cassandra M. Wilkinson, Jennifer I. Burrell, Jonathan Kuziek, Sibi Thirunavukkarasu, Brian Buck, Kyle E. Mathewson
Abstract
Abstract In this study, we demonstrated the use of low-cost portable electroencephalography (EEG) as a method for prehospital stroke diagnosis. We used a portable EEG system to record data from 25 participants, 16 had acute ischemic stroke events, and compared the results to age-matched controls that included stroke mimics. Delta/alpha ratio (DAR), (delta + theta)/(alpha + beta) ratio (DBATR) and pairwise-derived Brain Symmetry Index (pdBSI) were investigated, as well as head movement using the on-board accelerometer and gyroscope. We then used machine learning to distinguish between different subgroups. DAR and DBATR increased in ischemic stroke patients with increasing stroke severity (p = 0.0021, partial η 2 = 0.293; p = 0.01, partial η 2 = 0.234). Also, pdBSI decreased in low frequencies and increased in high frequencies in patients who had a stroke (p = 0.036, partial η 2 = 0.177). Using classification trees, we were able to distinguish moderate to severe stroke patients and from minor stroke and controls, with a 63% sensitivity, 86% specificity and accuracy of 76%. There are significant differences in DAR, DBATR, and pdBSI between patients with ischemic stroke when compared to controls, and these effects scale with severity. We have shown the utility of a low-cost portable EEG system to aid in patient triage and diagnosis as an early detection tool.