Prediction of Diabetes through Retinal Images Using Deep Neural Network

Table 1

Comparison chart of the proposed work.

Model	Class-label	Precision	Recall	F1-score	Accuracy (%)
Logistic regression [33]	0 (nondiabetic)	0.72	0.71	0.93	73
	1 (diabetic)	0.73	0.72	0.94
Random forest	0 (nondiabetic)	0.76	0.75	0.75	77.4
	1 (diabetic)	0.75	0.77	0.76
Proposed fine-tuned MLP	0 (nondiabetic)	0.86	0.85	0.85	86.6
	1 (diabetic)	0.86	0.88	0.87
Deep neural network (proposed)	0 (nondiabetic)	0.95	0.91	0.93	95.6
	1 (diabetic)	0.94	0.93	0.94