Early Precise Prediction and Severe Risk Stratification of Intrahepatic Cholestasis of Pregnancy: Advances and Future Directions in Clinical Translation from Traditional Models to Artificial Intelligence and Multi-Omics Technologies

Intrahepatic cholestasis of pregnancy (ICP) is a liver disorder unique to pregnancy, closely associated with severe adverse maternal and fetal outcomes such as preterm birth and intrauterine fetal death. Its pathogenesis involves a complex interplay of genetic, hormonal, metabolic, and environmental factors, with a higher risk observed in southern China and among individuals co-infected with hepatitis B virus. Substantial evidence demonstrates a significant dose–response relationship between serum total bile acid (TBA) levels and perinatal outcomes, with TBA ≥ 40 μmol/L commonly used as a criterion for severe ICP. However, most existing prediction models are based on single-center, retrospective studies with small sample sizes and insufficient external validation, limiting their clinical generalizability. In recent years, nomograms and machine learning methods have demonstrated advantages in the early prediction and risk stratification of ICP. Deep learning models and multi-omics integration strategies have further enhanced predictive accuracy. Nevertheless, challenges remain regarding model interpretability, data standardization, and cross-population applicability. Future research should leverage large-scale, multi-center prospective cohorts, integrating multi-omics technologies, including genomics, metabolomics, gut microbiome profiling with artificial intelligence to develop clinically actionable and interpretable risk prediction tools. Integrating these tools into electronic health record systems and mobile platforms may facilitate the early identification and individualized management of ICP, ultimately improving maternal and neonatal outcomes.