Eye tracking for classification of concussion in adults and pediatrics.

Introduction: In order to obtain FDA Marketing Authorization for aid in the diagnosis of concussion, an eye tracking study in an intended use population was conducted.
Methods: Potentially concussed subjects recruited in emergency department and concussion clinic settings prospectively underwent eye tracking and a subset of the Sport Concussion Assessment Tool 3 at 6 sites. The results of an eye tracking-based classifier model were then validated against a pre-specified algorithm with a cutoff for concussed vs. non-concussed. The sensitivity and specificity of eye tracking were calculated after plotting of the receiver operating characteristic curve and calculation of the AUC (area under curve).
Results: When concussion is defined by SCAT3 subsets, the sensitivity and specificity of an eye tracking algorithm was 80.4 and 66.1%, The AUC was 0.718. The misclassification rate ( n = 282) was 31.6%.
Conclusion: A pre-specified algorithm and cutoff for diagnosis of concussion vs. non-concussion has a sensitivity and specificity that is useful as a baseline-free aid in diagnosis of concussion. Eye tracking has potential to serve as an objective "gold-standard" for detection of neurophysiologic disruption due to brain injury.
Competing Interests: Author US has an equity interest in the technology investigated in this paper via ownership of intellectual property assigned to NYU, VA, and HCMC and licensed to Oculogica Inc. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
(Copyright © 2022 Samadani, Spinner, Dynkowski, Kirelik, Schaaf, Wall and Huang.)