Eye movement dysfunction for screening mild cognitive impairment

Abstract Background Early screening of cognitive impairment is crucial to delay cognitive deterioration and daily living disability in the elderly. However, the complexity and high cost of existing strategies appear to be great limits to the early screening on a large scale. Eye movement test shows a huge advantage in the early identification of mild cognitive impairment (MCI) because of quick, non‐invasive and cost‐effective features. Method The study involved a total of 151 participants: 90 MCI patients and 50 cognitive normal controls (CN) . All participants underwent eye movement signals under four visual tasks, including direct gaze towards target, smooth pursuit tasks, saccading toward a jumping target and looking away a jumping target. Five features were extracted from eye movement signals by linear and nonlinear analysis, including saccade number in resting state, saccade number in smooth persuit tasks, saccade lantency, antisaccade lantency, antisaccade error rate. Discriminant features selection was performed using the logistic regression (LR) analysis. Receiver operator characteristic (ROC) analyses and the area under the curve (AUC) were employed to investigate classification ability. Result After the penalization of LR analysis, there were two top features that exhibited significant differences between MCI and CN, which included saccade number in smooth pursuit tasks and antisaccade error rate (p<0.05). ROC curves showed a favorable diagnostic accuracy of the two features in the discrimination between controls and MCI (AUC: 0.823, P<0.01). Conclusion These results suggest that eye movement test has the potential to screen patients with MCI from the elderly with cognitive normality. Therefore, eye movement test is a powerful tool for broad screening in community‐dwelling older adults.