Machine learning-based nomogram for mortality risk stratification in cirrhotic patients with sepsis: a single-center...

Area of Science:

• Hepatology
• Critical Care Medicine
• Medical Informatics

Background:

• Sepsis is a life-threatening complication in liver cirrhosis patients.
• Predicting in-hospital mortality in this population is crucial for timely intervention.
• Existing predictive models may lack accuracy or clinical utility.

Purpose of the Study:

• To develop and validate a nomogram-based predictive model for in-hospital mortality in liver cirrhosis patients with sepsis.
• To evaluate the model's predictive accuracy and clinical usefulness.

Main Methods:

• Retrospective collection of clinical data from 264 liver cirrhosis patients with sepsis.
• Development of a predictive model using Lasso regression on a training set (70%) and validation on a separate set (30%).
• Evaluation of the nomogram's performance using ROC curves, calibration plots, and decision curve analysis (DCA).

Main Results:

• The nomogram identified alcoholic cirrhosis, Child-Pugh score, mechanical ventilation, TBiL, and HR as independent predictors of mortality.
• The model demonstrated strong predictive performance with AUCs of 0.81 (training) and 0.83 (validation).
• Calibration plots showed good agreement between predicted and observed mortality, and DCA indicated significant clinical net benefit.

Conclusions:

• The developed nomogram model shows promising predictive potential for in-hospital mortality in this patient cohort.
• This tool can support clinical decision-making and guide early interventions for high-risk patients.
• Further validation in diverse clinical settings is warranted to confirm its broad applicability.