Can Diabetic Eye Testing Be Simplified?
Three-test models matched nine-test battery in classifying disease stages
-
By
-
Julie Greenbaum
-
February 18, 2026
-
Clinical Scorecard: Can Diabetic Eye Testing Be Simplified?
At a Glance
| Category | Detail |
|---|
| Condition | Diabetic Eye Disease |
| Key Mechanisms | Machine learning models classify stages of diabetic eye disease using age, sex, and visual function tests. |
| Target Population | Individuals with diabetes mellitus, primarily aged 60-67. |
| Care Setting | Diabetes clinics and sensory aging studies. |
Key Highlights
- Machine learning models achieved AUC values of 0.94 or higher for classifying diabetic eye disease stages.
- Top models utilized combinations of up to three visual function tests.
- Distance visual acuity and reading index were frequently included in high-performing models.
Guideline-Based Recommendations
Diagnosis
- Use machine learning models incorporating age, sex, and visual function tests for classification.
Management
- Consider multimodal retinal imaging and pharmacologic dilation in assessments.
Monitoring & Follow-up
- Longitudinal studies are needed to assess predictive capabilities of visual function measurements.
Risks
- Incomplete data due to testing limitations may affect model accuracy.
Patient & Prescribing Data
Participants from the Northern Ireland Sensory Ageing Study and diabetes clinics.
Models showed similar performance to traditional nine-test batteries, suggesting simplification is feasible.
Clinical Best Practices
- Incorporate machine learning models in routine diabetic eye disease assessments.
- Utilize a combination of visual function tests for optimal classification.
References
