Artificial Intelligence Improves Patient Follow-Up in a Diabetic Retinopathy Screening Program

Area of Science:

• Ophthalmology
• Medical Informatics
• Artificial Intelligence in Healthcare

Background:

• Diabetic retinopathy (DR) screening is crucial for preventing vision loss in diabetic patients.
• Traditional DR screening workflows can have delays in results and follow-up.
• The integration of Artificial Intelligence (AI) offers potential improvements in healthcare delivery.

Purpose of the Study:

• To compare the follow-up rates and reasons for follow-up in an AI-based DR screening workflow versus traditional human-based workflows.
• To evaluate the impact of an AI workflow on patient adherence to ophthalmology referrals after positive DR screening.

Main Methods:

• A DR screening program screened 2243 adult patients with diabetes.
• Follow-up rates were compared between an AI workflow (IDx-DR), a human workflow, and a hybrid AI-Human workflow for patients screening positive for more-than-mild DR (MTMDR).
• Patient follow-up decisions were surveyed via phone.

Main Results:

• The AI Workflow demonstrated a significantly higher ophthalmology follow-up rate (35.5%) compared to the Human Workflow (12.0%) and AI-Human Hybrid Workflow (11.7%).
• Under the AI Workflow, 69.2% of patients with MTMDR positive results followed up with an ophthalmologist within 90 days.
• The AI Workflow provided results within 48 hours, while other workflows took up to 7 days.

Conclusions:

• Ophthalmology follow-up rates after positive DR screening are approximately threefold higher with the AI Workflow compared to human-based workflows.
• The accelerated turnaround time for results in the AI Workflow may contribute to improved patient follow-up behavior.
• AI-driven screening holds promise for enhancing patient adherence in DR management.