To develop and validate a nomogram for individualized risk prediction and stratification of rapidly progressive diabetic retinopathy (PDR) in type 2 diabetes mellitus (T2DM).
Approach:
Key Findings:
Univariate analysis identified six significant factors (all P < 0.05): diabetes duration, HbA1c, 24-hour urinary protein quantification, GDF15, DRSS grade, and foveal avascular zone area.
Random Forest model achieved the highest validation AUC of 0.780 compared to Gradient Boosting Machine (0.741) and multivariable logistic regression (0.698).
Calibration curves showed good consistency between predicted and observed probabilities.
DCA indicated high clinical net benefit of the model at 0.1-0.8 threshold probability.
Interpretation:
A novel risk prediction model for rapidly progressive PDR in T2DM was developed, demonstrating favorable discrimination and calibration.
Limitations:
Retrospective design may introduce selection bias.
Generalizability may be limited to similar patient populations.
Conclusion:
The model provides a tool for early identification of high-risk individuals.
