Screening of key variables and development and validation of a prognostic model for hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is a common primary liver cancer, and managing severe cases requiring ICU admission is challenging due to the lack of targeted prognostic tools. This study leveraged the MIMIC-IV database to develop a prognostic model for critically ill HCC patients. A retrospective cohort was randomly split into training (70%) and validation (30%) sets. Through sequential baseline analysis, univariate Cox regression, LASSO regression, and multivariate Cox regression, five key variables were identified: glucose, anion gap, Acute Physiology Score III (APS III), percutaneous oxygen saturation (SpO₂), and congestive heart failure. The prognostic model demonstrated excellent discriminative ability, generalizability, and predictive accuracy in both datasets. A nomogram incorporating these variables was developed, confirming its clinical utility. This tool may assist clinicians in identifying high-risk HCC patients and formulating individualized treatment plans for this vulnerable population.