Table 2.

Performance of algorithms with missing data handled by mean imputation.

Algorithm	Mean imputation
	AUCa, mean (SD)	P value	APb, mean (SD)	P value	Sensitivity, mean (SD)	P value	Specificity, mean (SD)	P value
LRc	0.82 (0.04)	.02	0.87 (0.03)	.047	0.50 (0.10)	<.001	0.91 (0.03)	>.99
LR+CSLd	0.82 (0.04)	.02	0.87 (0.03)	.03	0.67 (0.07)	.002	0.82 (0.02)	.98
DTe	0.65 (0.05)	<.001	0.76 (0.03)	<.001	0.43 (0.09)	<.001	0.87 (0.03)	>.99
DT+CSL	0.64 (0.02)	<.001	0.75 (0.03)	<.001	0.41 (0.05)	<.001	0.87 (0.02)	>.99
RFf	0.84 (0.05)	.52	0.89 (0.03)	.90	0.01 (0.01)	<.001	1.00 (0.00)	>.99
RF+CSL	0.84 (0.05)	.67	0.89 (0.03)	.93	0.64 (0.09)	.001	0.84 (0.03)	>.99
SVMg (RBFh)	0.78 (0.06)	<.001	0.85 (0.03)	<.001	0.12 (0.04)	<.001	0.99 (0.01)	>.99
SVM (RBF)+CSL	0.81 (0.05)	<.001	0.86 (0.03)	<.001	0.76 (0.08)	.98	0.73 (0.03)	<.001
SVM (polyi)	0.74 (0.06)	<.001	0.83 (0.03)	<.001	0.50 (0.07)	<.001	0.84 (0.03)	>.99
SVM (poly)+CSL	0.81 (0.05)	<.001	0.87 (0.03)	<.001	0.77 (0.08)	.99	0.73 (0.03)	<.001
SVM (linear)	0.79 (0.04)	<.001	0.85 (0.03)	<.001	0.48 (0.07)	<.001	0.89 (0.02)	>.99
SVM (linear)+CSL	0.80 (0.04)	.004	0.86 (0.03)	.005	0.65 (0.07)	<.001	0.81 (0.02)	.94
SNNj	0.81 (0.05)	<.001	0.87 (0.03)	.003	0.32 (0.09)	<.001	0.95 (0.01)	>.99
SNN+CSL	0.81 (0.05)	<.001	0.87 (0.03)	<.001	0.65 (0.11)	.002	0.83 (0.02)	>.99
DNNk	0.83 (0.05)	.045	0.88 (0.03)	.13	0.33 (0.09)	<.001	0.96 (0.02)	>.99
DNN+CSL	0.84 (0.04)	N/Al	0.88 (0.03)	N/A	0.73 (0.09)	N/A	0.80 (0.03)	N/A

^aAUC: area under the receiver operating characteristic curve.

^bAP: average precision.

^cLR: logistic regression.

^dCSL: cost-sensitive learning.

^eDT: decision tree.

^fRF: random forest.

^gSVM: support vector machine.

^hRBF: radial basis function kernel.

ⁱpoly: polynomial kernel.

^jSNN: single neural network.

^kDNN: dual neural network.

^lN/A: not applicable.