Table 2.

Performance of different algorithms trained on the testing data set.

Models		Precision, %	Sensitivity, %	Specificity, %	F1 score	Balanced accuracy, %	AUCa (95% CI)	Accuracy, %
Model 1
	Adaptive boosting	43.34	55.37	89.29	0.4862	72.33	0.81 (0.77-0.82)	84.92
	Decision tree	68.61	35.47	97.55	0.4676	66.51	0.78 (0.76-0.80)	89.41
	Extra trees	78.56	59.95	97.53	0.6800	78.74	0.83 (0.77-0.83)	92.60
	Gradient boosting	52.58	49.35	93.29	0.5091	71.32	0.82 (0.77-0.83)	87.53
	k-nearest neighbor	47.32	50.92	91.45	0.4905	71.18	0.76 (0.76-0.84)	86.14
	Linear discriminant analysis	14.02	82.46	23.74	0.2397	53.10	0.75 (0.74-0.84)	31.43
	Light gradient boosting	91.76	62.70	99.15	0.7449	80.92	0.82 (0.77-0.83)	94.37
	Logistic regression	14.16	85.47	21.84	0.2430	53.66	0.68 (0.68-0.85)	30.18
	Multiple layers perception	16.64	78.80	40.44	0.2748	59.62	0.80 (0.68-0.85)	45.47
	Random forest	90.62	40.45	99.37	0.5593	69.91	0.81 (0.78-0.83)	91.64
	Extreme gradient boosting	79.34	71.86	97.18	0.7541	84.52	0.83 (0.78-0.84)	93.86
Model 2
	Adaptive boosting	61.83	72.33	93.45	0.6667	82.89	0.83 (0.80-0.84)	90.75
	Decision tree	78.50	63.52	97.45	0.7022	80.48	0.81 (0.80-0.82)	93.11
	Extra trees	74.48	60.59	96.96	0.6682	78.77	0.84 (0.80-0.85)	92.30
	Gradient boosting	83.08	67.92	97.97	0.7474	82.95	0.84 (0.82-0.85)	94.13
	k-nearest neighbor	87.37	52.20	98.89	0.6535	75.55	0.82 (0.81-0.86)	92.92
	Linear discriminant analysis	16.33	82.81	37.76	0.2728	60.29	0.76 ()0.76-0.86	43.52
	Light gradient boosting	77.97	75.68	96.86	0.7681	86.27	0.85 (0.80-0.85)	94.15
	Logistic regression	16.12	81.76	37.58	0.2692	59.67	0.73 (0.73-0.86)	43.23
	Multiple layers perception	16.19	80.08	39.21	0.2694	59.65	0.71 (0.71-0.86)	44.44
	Random forest	66.67	70.02	94.86	0.6830	82.44	0.82 (0.80-0.85)	91.69
	Extreme gradient boosting	78.95	78.62	96.92	0.7878	87.77	0.85 (0.81-0.86)	94.58

^aAUC: area under the curve.