Machine-learning algorithms based on personalized pathways for a novel predictive model for the diagnosis of hepatocellular carcinoma

Table 3 Performance evaluation of training and external validation sets

	Overall cross-study validation	External cohort validation
	GSE14520, GSE76427	TCGA-LIHC
AUPRC	0.981	0.967
AUROC	0.987(0.979–0.996)	0.992(0.985–0.998)
Brier score	0.091	0.112
ACC	0.967	0.868
Precision	0.976	1
Sensitivity (recall)	0.965	0.849
F1	0.97	0.918
MCC	0.934	0.639

ISSN: 1471-2105