Table 5.

Testing data results and comparison with other machine learning algorithms.

Model	TNa	FPb	FNc	TPd	Sene	Spef	Accg	BAh	AUCi
5-layer DNNj: copying	967	103	5	46	0.9020	0.9037	0.9037	0.9028	0.9617
5-layer DNN: SMOTEk [23]	984	86	8	43	0.8431	0.9196	0.9161	0.8814	0.9555
5-layer DNN with PCAl (8 features)	922	148	5	46	0.9020	0.8617	0.8635	0.8818	0.9549
Linear regression [24]	983	87	7	44	0.8627	0.9187	0.9161	0.8907	0.9563
Decision tree [25]	915	155	5	46	0.9020	0.8551	0.8573	0.8786	0.9252
Random forest [21]	955	115	5	46	0.9020	0.8925	0.8930	0.8972	0.9590
Support vector machine [26]	955	115	5	46	0.9020	0.8925	0.8930	0.8972	0.9588
XGBoostm [22]	945	125	6	45	0.8824	0.8832	0.8831	0.8828	0.9558
AdaBoost [19,20]	937	133	5	46	0.9020	0.8757	0.8769	0.8888	0.9586
GradBoost [27]	936	134	6	45	0.8824	0.8748	0.8751	0.8786	0.9525
HistBoost [28]	959	111	7	44	0.8627	0.8963	0.8947	0.8795	0.9535

^aTN: true negative.

^bFP: false positive.

^cFN: false negative.

^dTP: true positive.

^eSen: sensitivity.

^fSpe: specificity.

^gAcc: accuracy.

^hBA: balanced accuracy.

ⁱAUC: area under the curve.

^jDNN: deep neural network.

^kSMOTE: synthetic minority oversampling technique.

^lPCA: principal component analysis.

^mXGBoost: eXtreme Gradient Boosting.