Integrating Multi-Task Eye Tracking and Interpretable Machine Learning for High-Accuracy Screening of Amblyopia in...

Area of Science:

• Ophthalmology
• Neuroscience
• Developmental Pediatrics

Background:

• Amblyopia, or lazy eye, is a developmental vision disorder causing reduced acuity and contrast sensitivity.
• Current diagnostic methods may not fully capture the functional deficits in amblyopic vision.

Purpose of the Study:

• To develop and validate a brief, child-friendly eye-tracking protocol for assessing spatial vision in amblyopia.
• To identify objective oculomotor markers indicative of amblyopic visual dysfunction.

Main Methods:

• A battery of task-evoked eye-tracking paradigms was used to measure fixation stability, fine pattern processing, and smooth pursuit.
• Oculomotor traces were analyzed to derive markers like fixation dispersion, saccadic strategy, and pursuit gain.
• A machine learning classifier was trained using these markers for group discrimination.

Main Results:

• The eye-tracking protocol successfully differentiated between children with unilateral amblyopia and age-matched controls.
• Pursuit and orientation-dependent oculomotor markers were key features for distinguishing the groups.
• The method demonstrated stable performance and was well-tolerated in pediatric participants.

Conclusions:

• Task-evoked eye movements provide objective, complementary biomarkers for amblyopia assessment.
• This non-invasive protocol can aid in the screening and monitoring of amblyopia in pediatric settings.
• Further validation and age-stratified norms are needed for community and school-based vision care.