Application of artificial intelligence in chronic liver diseases: a systematic review and meta-analysis

Decharatanachart, Pakanat; Chaiteerakij, Roongruedee; Tiyarattanachai, Thodsawit; Treeprasertsuk, Sombat

doi:10.1186/s12876-020-01585-5

BMC Gastroenterology

Table 1 Characteristics of included studies

From: Application of artificial intelligence in chronic liver diseases: a systematic review and meta-analysis

Study/year	Country	Study cohort	Population	AI classifier	AI modality	Development cohort	Validation cohort	Validation methods	Stage	Sensitivity	Specificity	TP	FP	TN	FN
Zhang, 2012 [25]	China	Prospective	Chronic hepatitis B, C	ANN	USG	F1/F2/F3/F4 40/22/55/62	F1/F2/F3/F4 13/8/19/20	Independent test set	F4	0.95	0.85	19	6	34	1
Chen, 2017 [27]	China	Prospective	Chronic hepatitis B	NB^a, RF^a, KNN, SVM	Elastography	S0/S1/S2/S3/S4 119/164/88/72/70	N/A	k-fold cross validation	F4	0.6866	0.8854	48	51	392	22
									≥ F3	0.7866	0.8738	112	47	324	30
									≥ F2	0.7471	0.8621	172	39	244	58
									≥ F1	0.7967	0.8250	314	21	98	80
Choi, 2018 [29]	Korea	Retrospective	General population	CNN	CT	F0/F1/F2/F3/F4 3357/113/284/460/3247	F0/F1/F2/F3/F4 118/109/161/173/330	Independent test set	F4	0.846	0.966	279	19	242	51
									≥ F3	0.946	0.954	476	18	370	27
									≥ F2	0.955	0.899	634	23	204	30
Yasaka, 2018 [30]	Japan	Retrospective	General population	CNN	CT	F0/F1/F2/F3/F4 113/36/56/66/125	F0/F1/F2/F3/F4 29/9/14/16/32	Independent test set	F4	0.75	0.57	24	29	39	8
									≥ F3	0.75	0.65	36	18	34	12
									≥ F2	0.76	0.68	47	12	26	15
Yasaka, 2018 [32]	Japan	Retrospective	General population	CNN	MRI	F0/F1/F2/F3/F4 54/53/81/113/233	F0/F1/F2/F3/F4 10/10/15/20/45	Independent test set	F4	0.76	0.76	34	13	42	11
									≥ F3	0.78	0.74	51	9	26	14
									≥ F2	0.84	0.65	76	3	7	14
Li, 2019 [23]	China	Prospective	Chronic hepatitis B	Adaboost^a, DT, LR, ANN, RF, SVM	USG	F0/F1/F2/F3/F4 15/33/38/23/35	N/A	Tenfold cross validation	≥ F2	0.875	0.769	84	11	37	12
Wang, 2019 [28]	China	Prospective	Chronic hepatitis B	CNN	Elastography	F0-1/F2/F3/F4 43/72/85/66	F0-1/F2/F3/F4 22/37/41/32	Independent test set	F4	0.969	0.88	31	12	88	1
									≥ F3	0.904	0.983	66	1	58	7
									≥ F2	0.691	0.909	76	2	20	34
Ahmed, 2020 [31]	Egypt	Prospective	Chronic hepatitis C	SVM	MRI	22 fibrotic patients 15 healthy patients	N/A	Leave one out cross validation	≥ F1	0.818	0.866	18	2	13	4
Lee, 2020 [22]	Korea	Retrospective	Chronic liver disease, hepatitis B, C	CNN	USG	F0/F1/F23/F4 363/394/1652/1566	F0/F1/F23/F4 290/17/72/193	Independent test set	F4	0.778	0.937	150	24	355	43
Lee, 2020 [22]	Korea	Retrospective	Chronic liver disease, hepatitis B, C	CNN	USG	F0/F1/F23/F4 363/394/1652/1566	F0/F1/F23/F4 290/17/72/193	Independent test set	≥ F2	0.913	0.824	242	54	253	23
Schawkat, 2020 [40]	Switzerland	Prospective	General population	SVM	MRI	F0/F1/F2/F3/F4 5/7/13/8/8	F0/F1/F2/F3/F4 3/5/5/5/3	Independent test set	≥ F3	0.750	0.923	6	1	12	2
Piscaglia, 2006 [37]	Spain	Retrospective	Chronic hepatitis C	ANN	Clinical data	F0/F1/F3/F4 216/176/87/31	F3/total 23/96	Independent test set	≥ F3	0.783	0.890	18	8	65	5
Wang, 2010 [36]	China	Retrospective	Chronic hepatitis C	ANN	Clinical data	F0-1/F2-4 166/60	F0-1/F2-4 80/36	Independent test set	≥ F2	0.917	0.800	33	16	64	3
Raoufy, 2011 [39]	Iran	Prospective	Chronic hepatitis B	ANN	Clinical data	Cirrhotic/non-cirrhotic 11/75	Cirrhotic/non-cirrhotic 8/50	Independent test set	F4	0.875	0.920	7	4	46	1
Pournik, 2014 [35]	Iran	Retrospective	NAFLD patients	ANN	Clinical data	Cirrhotic/non-cirrhotic 52/248	Cirrhotic/non-cirrhotic 15/65	Independent test set	F4	0.66	0.99	44	4	309	23
Shousha, 2018 [34]	Egypt	Retrospective	Chronic hepatitis C	ANN^a, DT	Clinical data	F0-2/F3-4 204/223	N/A	k-fold cross validation	≥ F3	0.825	0.811	184	39	165	39
Wei, 2018 [33]	USA	Retrospective	Chronic hepatitis B, C	DT, RF, GB^a	Clinical data	S0/S1/S2/S3/S4 46/169/134/56/85	S0/S1/S2/S3/S4 15/21/12/11/27	Independent test set	S4	0.78	0.85	21	9	50	6
Wei, 2018 [33]	USA	Retrospective	Chronic hepatitis B, C	DT, RF, GB^a	Clinical data	S0/S1/S2/S3/S4 46/169/134/56/85	S0/S1/S2/S3/S4 15/21/12/11/27	Independent test set	≥ S3	0.84	0.85	32	7	41	6
Li, 2019 [38]	China	Retrospective	Chronic hepatitis B	DT^a, RF^a, LR, SVM	Clinical data	460 patients	460 patients	Independent test set	F4	0.596	0.705	56	108	258	38
									≥ F3	0.939	0.803	176	54	219	11
									≥ F2	0.970	0.763	319	31	100	10
Kuppili, 2017 [26]	Portugal	Prospective	Mixed population	ELM	USG	NAFLD/non-NAFLD 36/27 patients	N/A	K-fold cross validation		0.913	0.921	33	2	25	3
Byra, 2018 [24]	Poland	Prospective	Obese population	CNN	USG	NAFLD/non-NAFLD 38/17 patients	N/A	Leave one out cross validation		1.000	0.882	38	2	15	0

ANN artificial neural networks, CNN convolutional neural networks, NB Naïve Bayes, RF random forest, KNN k-nearest neighbor, SVM support vector machine, MLP multilayer perception, DT decision tree, GB gradient boosting, LR logistic regression, ELM extreme learning machine, F4 diagnosis of cirrhosis, ≥ F3 diagnosis of advanced fibrosis (F3–F4), ≥ F2 diagnosis of significant fibrosis (F2–F4)
^aSelected AIs in the analysis

Back to article page

ISSN: 1471-230X

Contact us

Submission enquiries: bmcgastroenterology@biomedcentral.com
General enquiries: ORSupport@springernature.com