New technology that tracks the eye movements of patients may be a more accurate measure of brain injury than any other diagnostic measurements currently in use, according to a study recently published in the journal Concussion. Dr. Uzma Samadani, who recently joined the faculty at Hennepin County Medical Center, in collaboration with researchers at the Steven and Alexandra Cohen Veterans Center at NYU Langone Medical Center, developed the technology that can serve as a biomarker for concussion by tracking patients’ eye movements as they watch music videos.
The eye tracking technology works by having patients watch a music video for 220 seconds while eye movements are measured using a tracking camera. Videos used in the study ranged from Disney’s Puss in Boots to Wavin Flag by K’Naan. Multiple measures of each eye’s movement, followed by comparisons of their positions over time are used to distinguish between normal subjects and those with concussion.
In the work, led by Uzma Samadani, MD PhD, Charles Marmar, MD, and Eugene Laska PhD, the investigators built a classifier based on 34 emergency room patients with brain injury and 34 uninjured healthy control subjects of similar age. A classifier is a mathematical model that converts a patient’s eye movement measures into a prediction of the concussive status of the individual. They then tested the models on a dataset of 255 subjects, of whom 8 had concussions, and found that the eye tracking test had an optimal sensitivity of 88% and specificity of 87%.
Typically, a classifier produces a score and a subject is classified as having a concussion if the score exceeds a predefined threshold value. The accuracy of a biomarker is measured by plotting the probability of a true versus false positivity at each possible threshold value and the Area Under the Curve (AUC) is computed. A perfect biomarker has an AUC of 1.00, while a worthless marker – no better than the chance toss of a coin – has an AUC of 0.50. Most tests used clinically have AUC’s greater than 0.80. For example, serum troponin, the most commonly performed blood test for heart attacks has an AUC ranging in various studies from 0.76 to 0.96. In this study, the eye tracking based classifier had an AUC of 0.88, and a cross-validated AUC of 0.85.
To read the rest of this article, published in Medical News.net, please click here.