Table 3:

Screening performance of similarity indices

Classifier Feature	TP	FP	TN	FN	Acc.	Sen.	Spe.	AUROC (95% CI)
SI_m3	20	17	43	21	0.624	0.488	0.717	0.73(0.62,0.82)
SI_m4	28	15	45	13	0.723	0.683	0.750	0.81(0.71,0.88)
SI_m5	28	15	45	13	0.723	0.683	0.750	0.78(0.69,0.86)
SI_m6	29	13	47	12	0.752	0.707	0.783	0.78(0.68,0.86)
SI_m7	31	10	50	10	0.802	0.756	0.833	0.79(0.69,0.87)
SI_m8	27	13	47	14	0.733	0.659	0.783	0.78(0.66,0.87)
SI_m9	25	13	47	16	0.713	0.610	0.783	0.77(0.64,0.85)
SI_m10	23	14	46	18	0.683	0.561	0.767	0.76(0.66,0.85)

SI_mi, i = 3,4, …,10: Similarity index of i-bit word, TP: true positive, FP: false positive, TN: true negative, FN: false negative, Acc.: accuracy, Sen.: sensitivity, Spe.: specificity, AUROC: area under ROC, CI: confidence interval