Table 4.

AI role in MASLD.

AI	Author	Type of study	Study duration	Participants (n)	Comparator arm	Recruitment	Outcome	Ref
ANNa	Liu C et al	Cross-sectional	-	American people (6,613)	-	Predicting the risk of NAFLD	NAFLD was observed when the estimated risk of its occurrence exceeded the calculated threshold (0.388) using the established ANN model.	(28)
Path AI’s NASH MLb	Ratziu V et al	Randomized, double-blind, placebo-controlled trial	72-week	NASH patient with confirmed biopsies (251) and fibrosis stage F1-F3	Evaluated histological features of NASH by pathologists	Evaluating histological features of NASH	The agreement between AI and pathologists ranged from 0.28 to 0.62 weighted kappa statistics across histological characteristics. Steatosis (0.46–0.62) had the highest agreement, while fibrosis, lobular inflammation, and hepatocyte ballooning had lower.	(29)

^a. ANN: artificial neural network, ^b. ML: machine-learning.