Electrooculography-Based Detection of Refractive Vision Problems.

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Signal Processing

Background:

• Early detection of visual impairments is difficult due to subtle symptoms.
• Current home-based vision diagnostics are often self-initiated and discrete.
• Refractive disorders are a common visual impairment requiring accurate monitoring.

Purpose of the Study:

• To investigate the feasibility of using electrooculography (EOG) for passive detection of refractive power changes.
• To develop and evaluate EOG-based classification models for refractive error detection.
• To assess the performance of EOG models in within-subject, temporal generalization, and across-subject scenarios.

Main Methods:

• Thirty-nine participants simulated refractive conditions using optometry trial lenses.
• Electrooculography (EOG) signals were recorded during visual tasks.
• Classification models were trained to predict simulated refractive power levels relative to baseline.

Main Results:

• Within-subject EOG models achieved high accuracy (0.950 ± 0.034) in within-condition tests.
• Temporal generalization of within-subject models showed highly variable and low overall accuracy (0.159 ± 0.285).
• Across-subject models significantly improved with fine-tuning (0.785 ± 0.123) compared to naive or normalized models.

Conclusions:

• EOG signals contain valuable information for classifying refractive power, especially in personalized monitoring.
• Generalizing EOG-based refractive error detection across different time points and individuals presents a substantial challenge.
• This study provides insights for developing passive, real-time EOG systems for visual condition monitoring.