Related Experiment Video
Updated: Mar 27, 2026
A Data-Driven Approach to Quantifying Immune States in Sepsis
Published on: February 7, 2025
Development and Validation of a 28-Day Mortality Prediction Model for Patients with Sepsis Complicated by Autoimmune
Zhiyang Wang1,2, Xin Xiao1, Shifeng Li1
1Department of Critical Care Medicine, The First Affiliated Hospital of Soochow University, Suzhou, 215006, People's Republic of China.
Objective:
This study aimed to develop a prognostic model for patients with sepsis complicated by autoimmune diseases using machine learning methods and validate the model.
Methods:
Data on patients with sepsis and autoimmune diseases were extracted from the MIMIC-IV database. Participants were randomly divided into training set and validation set according to the ratio of 7:3. The predictors were selected by using LASSO regression analysis and the Boruta algorithm which affect the 28-day prognosis of patients. A nomogram was developed based on independent risk factors identified by logistic regression for 28-day prognosis and was internally and externally validated using calibration curves and DCA. Based on nomogram scores, patients were stratified into high- and low-score groups, with KM analysis demonstrating significant differences in mortality between the cohorts.
Results:
A total of 1,481 patients from the MIMIC-IV database met inclusion criteria and an external validation set included 57 patients from the Department of Critical Care Medicine of the First Affiliated Hospital of Soochow University. Ten overlapping predictors (Age, Gender, BMI, WBC, BUN, PT, APTT, history of cerebrovascular disease, history of liver disease, and CRRT) were identified by LASSO and Boruta algorithms and were subsequently confirmed as statistically significant independent risk factors through logistic regression. The prediction model built by the ten predictors showed superior predictive performance compared to the SOFA score in training (AUC=0.772), internal validation (AUC=0.771), and external validation cohorts (AUC=0.787). Hosmer-Lemeshow tests and calibration curves indicated strong agreement between predicted outcomes and actual observations across all cohorts, and DCA suggested significant clinical utility. The KM curve shows that the mortality rate of the high-score group is significantly higher than that of the low-score group.
Conclusion:
A prognostic model for predicting 28-day mortality in sepsis patients with autoimmune diseases demonstrated robust predictive performance and clinical applicability upon internal and external validation.
