Table 2.

Performance of models on age-stratified dataset.

Age group	Metric	DenseNet_121	ResNet_50	EfficientNet	VGG_11	MobileNet_v2
Age (18–39)	Accuracy	0.9209	0.9298	0.9452	0.9313	0.9472
Age (40–59)	Accuracy	0.9211	0.8786	0.8649	0.9222	0.8624
Age (60–90)	Accuracy	0.9342	0.8668	0.9165	0.9212	0.9144
Across age groups p value for Accuracy		6.96E−230	0.386	1.31E−17	1.73E−69	1.418E−89
Age (18–39)	F1-score	0.7638	0.7033	0.8096	0.8102	0.8123
Age (40–59)	F1-score	0.9435	0.9133	0.9044	0.9476	0.9067
Age (60–90)	F1-score	0.9649	0.9181	0.9554	0.957	0.9542
Across age groups p value for F1-score		<0.001	1.27E−30	5.76E−99	2.49E−13	1.260E−135
Age (18–39)	MCC	0.7224	0.7039	0.786	0.7776	0.7893
Age (40–59)	MCC	0.8129	0.7115	0.6746	0.8019	0.6502
Age (60–90)	MCC	0.4333	0.4453	0.2925	0.4968	0.3116
Across age groups p value for MCC		3.65E−234	0.0014	0.0065	4.37E−59	1.173E−112