Table 2.

The accuracy, sensitivity, specificity, and Dice score of three models in predicting the severity of DR on internal validation and external test datasets

Model		DRForecastGAN model	CycleGAN model	Pix2Pix model
Internal validation datasets	Accuracy	0.82 (0.80, 0.85)	0.76 (0.73, 0.79)	0.68 (0.65, 0.71)
	Sensitivity	0.77 (0.75, 0.80)	0.56 (0.53, 0.59)	0.72 (0.69, 0.75)
	Specificity	0.85 (0.82, 0.87)	0.84 (0.81, 0.86)	0.66 (0.64, 0.69)
	Dice	0.52	0.35	0.40
External test datasets	Accuracy	0.75 (0.72, 0.77)	0.61 (0.58, 0.64)	0.66 (0.63, 0.69)
	Sensitivity	0.85 (0.83, 0.87)	0.77 (0.74, 0.79)	0.58 (0.56, 0.61)
	Specificity	0.71 (0.68, 0.73)	0.55 (0.52, 0.58)	0.69 (0.66, 0.71)
	Dice	0.46	0.31	0.38