The acute-on-chronic liver failure (ACLF) classification, proposed by the World Gastroenterology Organisation (WGO), attempts to cover all ACLF patients diagnosed in the East and West. This study aimed to explore and establish a prognostic model based on this classification.

A total of 1159 hepatitis B virus–ACLF patients, enrolled with 90-day follow-up data, were divided into three groups (type A, B, and C) according to WGO ACLF classification and analyzed. A model of ACLF prognosis based on type (MAPT) was developed in a derivation cohort (n = 566); its reproducibility was tested in a validation cohort (n = 593).

A significant difference in 90-day mortality among the three groups was observed (31.1%, type A; 40.9%, type B; 61.4%, type C, P < 0.001). ACLF type was determined to be an independent risk factor of 90-day mortality in HBV-ACLF patients. An MAPT, inclusive of type and five other variables, was built and validated; it was found to be superior to the Chronic Liver Failure (CLIF) Consortium ACLF score, Model for End-Stage Liver Disease, CLIF-Sequential Organ Failure, and Child-Turcotte-Pugh scores in predicting 90-day mortality, with an area under the receiver operating characteristic curve of 0.802 (95% CI [0.763–0.836]), sensitivity of 71.77%, and specificity of 75.82%.

The MAPT model showed excellent predictive value for 90-day mortality in HBV-ACLF and can likely expand the clinical application of WGO ACLF classification.