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

## TL;DR

This study creates a machine learning model to predict in-hospital mortality in cirrhotic patients with sepsis, which could help guide clinical decisions.

## Contribution

A novel nomogram-based predictive model using Lasso regression for mortality risk stratification in cirrhotic sepsis patients.

## Key findings

- The model achieved an AUC of 0.81 in training and 0.83 in validation sets.
- Calibration plots showed predictions aligned closely with actual outcomes.
- The model provided significant clinical net benefit across threshold probabilities.

## Abstract

To develop and validate a nomogram-based predictive model for in-hospital mortality among patients with liver cirrhosis complicated by sepsis, and to evaluate its predictive accuracy.

Clinical data were retrospectively collected from patients diagnosed with liver cirrhosis and sepsis who were admitted to the Fuyang Infectious Disease Clinical College of Anhui Medical University between January 2018 and July 2025. Patients were classified into the Survivor group or the Non-survivor group. The dataset was randomly divided into a training set (70%) and a validation set (30%). Potential predictors were identified through univariate and multivariate logistic regression analyses, and a predictive model was subsequently developed using Lasso regression. The model was visualized as a nomogram, and its performance was rigorously evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA) to assess its clinical utility.

A total of 264 patients were enrolled in this study. Among the 188 patients in the training set, 54 (28.7%) died during hospitalization, while 21 out of 76 patients (27.6%) in the validation set experienced in-hospital mortality. Multivariate logistic regression analysis identified alcoholic cirrhosis, Child-Pugh score, mechanical ventilation, TBiL and HR as independent predictors of in-hospital mortality (all P < 0.05). The nomogram model demonstrated robust predictive performance, with ROC analysis showing an area under the curve (AUC) of 0.81 (95% CI: 0.75–0.81) in the training set and 0.83 (95% CI: 0.73–0.92) in the validation set. Calibration plots revealed that the model's predictions closely aligned with the ideal reference line. DCA showed that the model provided significant clinical net benefit across a wide range of threshold probabilities.

The nomogram model developed using Lasso regression appears to demonstrate promising predictive potential for in-hospital mortality in patients with liver cirrhosis complicated by sepsis. This tool may offer valuable support for clinical decision-making and could potentially aid in guiding early interventions for patients identified as higher risk.