Prognostic risk classification of metabolic dysfunction-associated fatty liver disease： Data-driven exploration and prospect

Metabolic dysfunction-associated fatty liver disease （MAFLD）， as one of the most common chronic liver diseases in the world， poses a severe challenge to precision diagnosis and treatment due to its complex pathogenesis and highly heterogeneous disease progression. Existing clinical classification systems cannot meet the needs for comprehensively analyzing the complexity of the disease and the heterogeneity of its adverse outcomes. In recent years， data-driven prognostic risk classification methods have gradually emerged， optimizing the ability for predicting adverse outcomes and enhancing the accuracy of identifying different endpoint outcomes. However， such paradigm of “classify first， associate outcomes later” suffers from a “black-box” nature， and there are various indicators for classification， leading to limited stability and generalizability in clinical application. Future research needs to integrate or establish large-scale population cohorts， develop outcome-oriented prognostic risk classification models， incorporate dynamic data， refine classification algorithms， and validate their generalizability across multiple populations， thereby providing reliable support for the precision diagnosis and treatment of MAFLD.