Table 2. Results from CP Models at the Significance Level 0.3^a.

descriptors	validation set	validity steatotic class	validity nonsteatotic class	efficiency	sensitivity	specificity	BA	MCC
RDKit	TR	0.73	0.72	0.69	0.62	0.59	0.61	0.14
	TS	0.74	0.71	0.75	0.67	0.61	0.64	0.18
5 selected RDKit	TR	0.77	0.73	0.71	0.67	0.61	0.64	0.18
	TS	0.76	0.79	0.71	0.67	0.7	0.69	0.24
ToxPrint	TR	0.77	0.71	0.71	0.67	0.59	0.63	0.17
	TS	0.71	0.69	0.74	0.63	0.57	0.6	0.13
transporter prediction	TR	0.86	0.72	0.41	0.76	0.28	0.52	0.03
	TS	0.88	0.76	0.44	0.82	0.43	0.62	0.18
NR prediction	TR	0.76	0.72	0.52	0.57	0.46	0.52	0.02
	TS	0.79	0.74	0.5	0.61	0.48	0.54	0.06
RDKit + ToxPrint	TR	0.78	0.73	0.69	0.68	0.61	0.64	0.19
	TS	0.74	0.71	0.82	0.71	0.65	0.68	0.24
RDKit + transporter prediction	TR	0.73	0.73	0.69	0.63	0.6	0.62	0.16
	TS	0.74	0.71	0.77	0.69	0.62	0.65	0.2
RDKit + NR prediction	TR	0.74	0.72	0.69	0.65	0.6	0.62	0.16
	TS	0.76	0.74	0.73	0.71	0.63	0.67	0.23
ToxPrint + transporter prediction	TR	0.78	0.71	0.72	0.7	0.6	0.65	0.2
	TS	0.74	0.69	0.77	0.67	0.6	0.63	0.17
ToxPrint + NR prediction	TR	0.8	0.73	0.69	0.71	0.61	0.66	0.2
	TS	0.68	0.71	0.76	0.61	0.62	0.61	0.15
transporter prediction + NR prediction	TR	0.8	0.72	0.54	0.61	0.48	0.55	0.06
	TS	0.79	0.69	0.6	0.73	0.47	0.6	0.14
RDKit + ToxPrint + transporter prediction	TR	0.81	0.71	0.69	0.72	0.58	0.65	0.2
	TS	0.74	0.72	0.82	0.71	0.65	0.68	0.24
RDKit + transporter prediction + NR prediction	TR	0.77	0.74	0.67	0.67	0.61	0.64	0.18
	TS	0.71	0.73	0.76	0.67	0.64	0.65	0.21
ToxPrint + transporter prediction + NR prediction	TR	0.79	0.71	0.71	0.71	0.59	0.65	0.20
	TS	0.71	0.70	0.80	0.66	0.62	0.64	0.18
RDKit + ToxPrint + NR prediction	TR	0.76	0.71	0.73	0.66	0.61	0.63	0.17
	TS	0.74	0.74	0.79	0.69	0.67	0.68	0.24
RDKit + ToxPrint + transporter prediction + NR prediction	TR	0.78	0.72	0.7	0.65	0.61	0.63	0.17
	TS	0.74	0.73	0.8	0.7	0.65	0.68	0.23

^aPerformances on training set (TR) and test set (TS) are shown. BA is balanced accuracy, and MCC is Matthews correlation coefficient.