A Diagnostic Model for Screening Diabetic Retinopathy Using the Hand-Held Electroretinogram Device RETeval

Area of Science:

• Ophthalmology
• Medical Technology
• Data Science

Background:

• Diabetic retinopathy (DR) is a leading cause of vision loss in diabetic patients.
• Early detection and screening are crucial for managing DR and preventing vision impairment.
• Current screening methods may have limitations in sensitivity and specificity.

Purpose of the Study:

• To develop and validate a novel screening model for diabetic retinopathy (DR) using the RETeval device.
• To assess the diagnostic performance of a DR risk score and a decision tree model for DR detection.

Main Methods:

• A cross-sectional study involving 232 diabetic patients and 70 controls.
• Utilized the RETeval device to obtain a DR risk score for each subject.
• Employed Receiver Operating Characteristic (ROC) curve analysis to determine optimal diagnostic cutoffs.
• Constructed Random Forest and Decision Tree models for DR diagnosis.

Main Results:

• The DR risk score showed a gradual increase with DR severity.
• The ROC curve analysis for the DR score yielded an Area Under the Curve (AUC) of 0.881, with optimal cutoff at 22.95 (sensitivity 74.3%, specificity 90.6%).
• A decision tree model, using the top four risk factors, achieved higher sensitivity (93.3%) and specificity (80.3%) for DR diagnosis.

Conclusions:

• The combination of the DR risk assessment protocol and decision tree model offers an innovative approach to DR risk evaluation.
• This integrated method enhances diagnostic sensitivity, making it a potentially superior alternative to current DR screening protocols.
• The developed model shows promise for improving the early detection and management of diabetic retinopathy.