Table 2.

Prediction performance of all nine predictors in the independent dataset.

	DisEMBL	IUPred	VSL2	Espritz	PONDR-FIT	MFDp2	IUPred2	AUCpreD	This Work
Sens	0.454	0.656	0.82	0.529	0.728	0.78	0.647	0.609	0.811 ± 0.007
Spec	0.915	0.872	0.735	0.932	0.856	0.857	0.87	0.908	0.856 ± 0.006
Acc	0.784	0.811	0.759	0.818	0.82	0.835	0.811	0.823	0.844 ± 0.003
Acc-b	0.684	0.764	0.777	0.731	0.792	0.819	0.761	0.759	0.834 ± 0.001
MCC	0.424	0.532	0.507	0.521	0.569	0.615	0.53	0.53	0.639 ± 0.003
F1	0.544	0.663	0.659	0.622	0.696	0.729	0.66	0.66	0.747 ± 0.002
AUC_ROC	0.779	0.824	0.841	0.888	0.857	0.88	0.822	0.872	0.9 ± 0.002
AUC_PR	0.617	0.673	0.642	0.754	0.695	0.622	0.672	0.72	0.789 ± 0.005

Note Bene. The new strategy was optimized five times under five-fold cross-validation. Therefore, the performance was also tested in the independent test dataset five times. The results shown in the table is the average of all five times. The highest value in each of these measures is in bold and highlighted (red).