Application of artificial intelligence in hepatology: Minireview

Table 1 Clinical applications of artificial intelligence

	Samples	Diagnosis	AI technique	Accuracy, %	AUROC	Ref.
Focal liver disease detection	US	Benign tumors	DL	97.2	NA	[23]
	Serum tests, clinical data	HCC	ML (gradient boosting)/DL	87.34/83.54	0.940/0.884	[25]
Diffuse liver disease staging	US	FLD	DL/SVM/ELM	100/82/92	1.0/0.79/0.92	[26]
	US	NAFLD	DL	NA	0.9777	[27]
	Elastography	Cirrhosis	DL	NA	0.97	[32]
Risk assessment	Clinical, pathohistological data	Poorer survival after HCC resection	2 DL models	NA	0.78, 0.75 (c-index)	[35]
	Sequence data	Poorer survival after HCC resection	DL	NA	0.68 (c-index)	[36]
	Clinical data	HCC development	ML	NA	0.64 (c-index)	[37]
	Clinical, histological data	1-yr and 3-yr clinical outcomes	ML	NA	0.78, 0.76	[38]

AUROC: Area under the receiver operating characteristics; c-index: Confidence interval; DL: Deep learning; ELM: Extreme learning machine; FLD: Fatty liver disease; HCC: Hepatocellular carcinoma; ML: Machine learning; NAFLD: Nonalcoholic fatty liver disease; SVM: Support vector machine; US: Ultrasonography; NA: Not available.