Table 4.

Performance comparisons between Meta-iAVP and the six machine learning algorithms as assessed by the five-repeated five-fold cross-validation and independent validation tests.

Dataset	Method a	Ac (%)	Sn (%)	Sp (%)	MCC
T544p+407n	k-NN	78.79	88.24	66.13	0.56
	rpart	74.09	81.03	64.82	0.47
	glm	70.15	82.87	53.27	0.38
	RF	84.22	85.70	82.34	0.68
	XGBoost	84.33	86.69	80.97	0.68
	SVM	79.53	83.81	73.86	0.58
	Meta-predictor	88.17	89.23	86.94	0.76
T544p+544n	k-NN	84.15	82.53	86.07	0.68
	rpart	80.63	82.37	79.73	0.62
	glm	77.11	77.78	76.78	0.54
	RF	89.44	84.18	94.68	0.79
	XGBoost	89.16	87.48	90.90	0.78
	SVM	88.79	87.13	90.71	0.78
	Meta-predictor	92.31	88.44	96.16	0.85
V60p+45n	k-NN	80.77	95.00	61.36	0.61
	rpart	75.96	86.67	61.36	0.50
	glm	68.27	86.67	43.18	0.34
	RF	86.54	86.67	86.36	0.73
	XGBoost	83.65	85.00	81.82	0.67
	SVM	86.54	93.33	77.27	0.72
	Meta-predictor	95.19	96.67	93.18	0.90
V60p+60n	k-NN	89.83	85.00	94.83	0.80
	rpart	83.05	88.33	77.59	0.66
	glm	73.73	78.33	68.97	0.48
	RF	91.53	90.00	93.10	0.83
	XGBoost	90.68	90.00	91.38	0.81
	SVM	89.83	88.33	91.38	0.80
	Meta-predictor	94.92	93.33	96.55	0.90

^ak-NN: k-nearest neighbor, rpart: ecursive partitioning and regression trees, glm: Generalized linear model, RF: Random forest, XGBoost: Extreme gradient boosting, and SVM: Support vector machine.