. 2025 Aug 18;44(1):97–105. doi: 10.1007/s11604-025-01853-y

Table 2.

AI model performance

References	Supplementary information	Dataset	Sample size (test group)		Tests
References	Supplementary information	Dataset	HCC	non-HCC	Sensitivity	Specificity	Accuracy	AUROC	NPV	PPV
[20]		Internal	185	185	0.830	0.950		0.890
[21]	Age, gender, spatial morphology	Internal	63	57	0.946	0.885	0.917	0.958	0.938	0.903
[21]	Age, gender, spatial morphology	Internal	63	57	0.981	0.899	0.942	0.963	0.977	0.915
[22]	Tumour marking information	Internal	50	12	0.750	0.880		0.870
[22]		Internal	50	12	0.750	0.820		0.870
[23]		Internal	218	167	0.815	0.902	0.853	0.899	0.787	0.917
[23]		External	264	292	0.739	0.889	0.805	0.869	0.727	0.895
[24]	Age, gender, global liver information	Internal	252	632	0.853	0.833	0.847	0.920
[24]		External	140	452	0.829	0.872	0.863	0.936
[25]	Age, gender, pertinent medical history	Internal	752	2103	0.986	0.960		0.990
		External-1	106	1517	0.758	0.986		0.991
		External-2	106	1415	0.918	0.955		0.980
		External-3	117	1561	0.917	0.965		0.980
		External-4	64	1740	0.878	0.956		0.982
[26]		Internal	Not specified (random 25% of total FLL)	Not specified (random 25% of total FLL)	0.630	0.931		0.884