Diabetic Retinopathy (DR) nomogram construction based on optical coherence tomography angiography parameters: a preliminary exploration of DR prediction.

Abstract

Aims: To construct a diabetic retinopathy (DR) prediction nomogram based on optical coherence tomography angiography (OCTA) parameters. Ophthalmologists can then use this nomogram to assess the risk of early-stage DR.

Methods: In this retrospective study, patients with type 2 diabetes mellitus who completed DR screening were enrolled and divided into training and validation sets. Fifteen parameters, including OCTA parameters, axial length (AL), age, and sex, were selected via least absolute shrinkage and selection operator (LASSO) in the training set. The chosen parameters were used to construct the model. Model performance was evaluated for both the training and validation sets via receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). A corresponding nomogram was created.

Results: A total of 464 eyes from 464 patients were divided into a training set (324, 69.83%) and a validation set (140, 30.17%). The superficial parafoveal capillary density (CD), deep parafoveal CD, foveal CD in the 300 µm-wide area surrounding the foveal avascular zone (FD- 300 area), AL, and patient ages were included in the final model. The area under curve of the model was 0.825 in the training set and 0.831 in the validation set. The calibration curves showed good alignment between the actual and predicted outcomes in both datasets. DCA demonstrated that the nomogram was clinically useful.

Conclusions: A model with good performance for predicting DR via OCTA parameters was developed. The superficial parafoveal CD, deep parafoveal CD, and FD- 300 area were important predictive parameters in this model. The corresponding nomogram may serve as a convenient tool for early DR risk prediction and lay the foundation for developing OCTA-based automated diagnostic software for early DR detection.