Exposure-shaped immunometabolic networks in chronic liver disease: translating multi-omics and artificial...

Chronic liver disease develops through sustained interactions among metabolic stress, environmental exposure, immune activation, and tissue remodeling. These processes are often studied as separate domains, yet in patients they converge within the same hepatic microenvironment and help explain why disease trajectories vary even within the same diagnostic category. Exposure science has sharpened attention to diet, alcohol, pollutants, chemical mixtures, gut-derived signals, sleep and circadian disruption, and other behavioral determinants of liver injury. At the same time, multi-omics approaches now capture complementary dimensions of disease biology, including genetic susceptibility, epigenetic memory, inflammatory transcriptional programs, proteomic signaling, metabolic rewiring, microbiome composition, and spatially restricted cell states. The central challenge is no longer simply to generate more data, but to connect these layers into clinically useful markers of progression. In this review, we discuss chronic liver disease as a set of exposure-shaped immunometabolic network states that extend across steatosis, inflammation, fibrosis, cirrhosis, and hepatocellular transformation. We summarize how major exposure domains feed into shared pathogenic hubs, how immune and metabolic circuits sustain injury, and how genomics, epigenomics, transcriptomics, proteomics, metabolomics, microbiome profiling, and single-cell or spatial methods can reveal biologically coherent biomarker candidates. We also examine how machine learning, deep learning, network-based modeling, and causal-inference strategies may support risk stratification and progression forecasting when used with attention to interpretability, cohort structure, and external validation. A prevention-oriented biomarker framework should identify transition-prone states early enough to guide monitoring, referral, treatment selection, or lifestyle intervention. Such translation will require better exposure assessment, longitudinal sampling, assay standardization, and transportable models tested across real-world populations.