A 4-center randomized controlled trial in Taiwan is testing whether AI-assisted fundus imaging improves screening for diabetic retinopathy and age-related macular degeneration in adults with diabetes and adults aged 50 and older at risk for AMD, versus usual physician-only screening. The study protocol, published as a multicenter, 2-arm, open-label RCT, will measure detection rates, positive predictive value, and cost-effectiveness, with completion expected in December 2027.

Why It Matters To Your Practice

• Diabetic retinopathy and AMD remain leading causes of vision loss, but screening opportunities are often missed in primary care and geriatrics.

• This trial targets a real-world gap: limited ophthalmic screening expertise among non-eye specialists in family medicine settings.

• If effective, AI could help clinicians identify more patients needing ophthalmology referral earlier in the disease course.

Clinical Implications

• The intervention compares AI-assisted screening with physician-only screening after a pilot phase, using individual 1:1 randomization.

• Eligible patients include adults aged 20 and older with Diabetes mellitus (DM) for diabetic retinopathy screening and adults aged 50 and older for AMD screening.

• Primary outcomes are practical for frontline care: confirmed detection rates and positive predictive value among those who screen positive.

Insights

• The study is being conducted across Family Medicine and Geriatric/Gerontology care at 4 medical centers in Taiwan, which improves relevance to outpatient practice.

• It is designed to assess not just diagnostic performance, but also cost-effectiveness — a key question for health systems considering AI deployment.

• Because this is an open-label protocol study, no efficacy results are available yet; the value for clinicians is in watching whether AI improves workflow and case finding in routine care.

The Bottom Line

• This is a practice-facing RCT of AI in primary care eye screening, not a retrospective algorithm study.

• For clinicians, the key question is whether AI meaningfully increases accurate early detection of DR and AMD enough to justify implementation costs and referral impact.

• Results are pending, but this is the kind of trial that could shape how AI enters everyday screening workflows.