Detecting cognitive impairment by eye movement analysis using automatic classification algorithms.

Area of Science:

• Neuroscience
• Computer Science
• Machine Learning

Background:

• Mild cognitive impairment (MCI) is a precursor to Alzheimer's Disease (AD).
• The Visual Paired Comparison (VPC) task, using eye-tracking, shows potential for detecting memory impairments.
• Current VPC methods rely on novelty preference, with limited classification performance.

Purpose of the Study:

• To improve the accuracy of detecting MCI using machine learning applied to VPC eye-tracking data.
• To explore the utility of eye movement characteristics beyond novelty preference for MCI detection.

Main Methods:

• Applied machine learning algorithms, specifically Support Vector Machines (SVMs), to analyze eye movement data from the VPC task.
• Modeled features including fixations, saccades, and re-fixations.
• Compared classification performance against traditional novelty preference metrics.

Main Results:

• The machine learning approach achieved 87% accuracy, 97% sensitivity, and 77% specificity in distinguishing MCI subjects from controls.
• This significantly outperformed the 67% accuracy, 60% sensitivity, and 73% specificity obtained using only novelty preference.

Conclusions:

• Machine learning techniques applied to VPC eye-tracking data significantly enhance the detection of MCI.
• This approach offers a promising, more accurate method for identifying cognitive impairments, potentially aiding in early AD prediction.