Abstract: Objective To explore a predictive model for 90-day mortality in patients with Hepatitis B Virus-associated Acute-on-Chronic Liver Failure (HBV-ACLF), thereby assisting in early clinical decision-making for timely intervention and treatment.Methods 88 HBV-ACLF patients admitting who met the inclusion and exclusion criteria and were admitted to Xiangyang Central Hospital from January 2019 to November 2023 were involved in the retrospective analysis. Patients were divided into the survival group with 61 cases (69.32%) and the mortality group with 27 cases (30.68%) based on their 90-day outcomes. Clinical data of both groups were collected, and independent risk factors for 90-day outcomes were identified using multivariate logistic regression. A prognostic prediction model was established and internally validated using 1000 bootstrap resamplings to assess its reliability.Results Among the 88 HBV-ACLF patients, multivariate regression identified C-reactive protein (CRP), procalcitonin (PCT), the ratio of measured liver volume to estimated liver volume (LV/ELV%), and the COSSH-ACLF IIs as independent risk factors for 90-day mortality. A scoring model was developed as Logit(P) = 0.285 × ln(CRP) + 5.291 × ln(PCT) + 1.017 × ln(COSSH-ACLF IIs)-9.757 × ln(LV/ELV%)-7.175. The area under the receiver operating characteristic curve (AUROC) of this model was significantly higher than that of COSSH-ACLF IIs alone, as well as the scores combined with CRP and PCT, or the scores combined with LV/ELV% (P<0.001). Internal validation confirmed the reliability of the combined predictive model.Conclusion The combined model offers higher predictive efficacy for the 90-day prognosis of HBV-ACLF patients, providing evidence-based medical support for early clinical diagnosis and treatment.

Key words: Hepatitis B virus, Acute-on-Chronic Liver Failure, Predictive Model Construction, Model Validation