Referable Diabetic Retinopathy Prediction Algorithm Applied to a Population of 120,389 Type 2 Diabetics over 11 Years...

Area of Science:

• Ophthalmology
• Diabetology
• Medical Informatics

Background:

• Diabetic retinopathy (DR) screening is crucial for preventing vision loss in diabetes patients.
• Low patient attendance rates significantly hinder the effectiveness of current DR screening programs.
• Developing predictive tools can optimize screening strategies and improve patient outcomes.

Purpose of the Study:

• To evaluate an algorithm's efficacy in predicting the development of any type of DR.
• To assess the algorithm's ability to predict referable DR, requiring specialist attention.
• To demonstrate the potential for personalized DR screening plans.

Main Methods:

• A retrospective study analyzed data from 120,389 patients with type 2 diabetes mellitus (T2DM).
• An 11-year follow-up period was utilized to track DR development.
• The predictive performance of the developed algorithm was assessed using Area Under the Curve (AUC) metrics.

Main Results:

• The algorithm achieved an AUC of 0.93 (95% CI, 0.92-0.94) for predicting any DR.
• For referable DR, the algorithm demonstrated an AUC of 0.90 (95% CI, 0.89-0.91).
• These results indicate a high level of agreement and predictive accuracy.

Conclusions:

• The developed algorithm effectively predicts the likelihood of developing any DR and referable DR.
• This predictive capability supports the creation of individualized screening schedules for T2DM patients.
• Personalized screening plans can enhance resource allocation and patient management for DR.