Detecting cognitive impairment by eye movement analysis using automatic classification algorithms

► A novel application of automatic classification methods from computer science to improve the accuracy of detecting Mild Cognitive Impairment during the Visual Paired Comparison task. ► An effective representation of eye movement characteristics such as fixations, saccades, and re-fixations as features for automatic classification algorithms. ► Our techniques allow to automatically distinguish age-matched normal control subjects from MCI subjects with 87% accuracy, 96% sensitivity and 77% specificity, compared to the best classification performance of 67% accuracy, 60% sensitivity, and 73% with previous techniques over VPC data.