GADNN: a revolutionary hybrid deep learning neural network for age and sex determination utilizing cone beam computed tomography images of maxillary and frontal sinuses

Table 5 The results of evaluation of machine learning and deep neural network models in determining sex

Method	Sex Groups	Precision	Recall	F1-Score	Accuracy
Logistic Regression with SMOTE^a	Female	0.70	0.55	0.62	0.62
Logistic Regression with SMOTE^a	Male	0.55	0.70	0.62	0.62
Multi-layer Perceptron with SMOTE	Female	0.75	0.62	0.68	0.67
Multi-layer Perceptron with SMOTE	Male	061	0.74	0.67	0.67
Random Forest with SMOTE	Female	0.75	0.72	0.74	0.71
Random Forest with SMOTE	Male	0.67	0.70	0.68	0.71
Deep Learning without SMOTE	Female	0.62	0.71	0.67	0.68
Deep Learning without SMOTE	Male	0.75	0.67	0.71	0.68
Deep Learning with SMOTE	Female	0.79	0.81	0.80	0.78
Deep Learning with SMOTE	Male	0.79	0.76	0.78	0.78
GADNN^b with SMOTE	Female	0.89	0.86	0.88	0.86
GADNN^b with SMOTE	Male	0.83	0.87	0.85	0.86

^athe synthetic minority oversampling technique; ^bGenetic algorithm based deep neural network model

ISSN: 1471-2288