This study used data from the MIMIC-IV database (training group) and an external hospital dataset (clinical validation group) to develop and validate a prediction model for sepsis-associated acute liver injury (SALI). The goal was to identify key risk factors and create an accessible clinical tool.

Study Population: The analysis included 1109 sepsis patients from MIMIC and 122 from the external hospital.

Ferritin is Key: After extensive statistical and machine learning analyses (LASSO regression, Boruta algorithm, PSM), higher serum ferritin level was confirmed as an independent risk factor that significantly differed between the SALI and non-SALI groups.

Predictive Risk Factors: The final logistic regression analysis identified six independent risk factors for secondary acute liver injury in sepsis patients.

Model Performance: A nomogram prediction model incorporating these six features demonstrated excellent predictive ability in both the training (AUC 0.765) and external validation (AUC 0.773) groups.

A higher serum ferritin level is an independent risk factor for acute liver injury in sepsis patients. The developed prediction model, which includes ferritin and five other clinical features, is a validated, high-value tool that can aid in the early risk stratification of sepsis patients in the ICU.

Read More