Related Experiment Video
Updated: May 24, 2026
08:46
A Neonatal Imaging Model of Gram-Negative Bacterial Sepsis
Published on: August 12, 2020
An explainable machine learning model for early pediatric sepsis prediction using cytokine and routine laboratory
Shiyao Li1, Jiaojiao Zhang1, Xiujun Deng1
1Department of Laboratory Medicine, the Affiliated Hospital of North Sichuan Medical College; Department of Laboratory Medicine, North Sichuan Medical College; Translational Medicine Research Center, North Sichuan Medical College, Nanchong, China.
Pediatric Research
|May 22, 2026
Summary
This study created an interpretable machine learning model for early pediatric sepsis prediction using cytokine data and infection markers, achieving high accuracy. The model is now an online calculator for clinical use.
Area of Science:
- Pediatric critical care medicine
- Machine learning in healthcare
- Biomarker discovery
Background:
- Early sepsis diagnosis in children is difficult due to vague symptoms.
- Developing accurate predictive models is crucial for timely intervention.
- Machine learning offers potential for improved diagnostic capabilities.
Purpose of the Study:
- To develop an interpretable machine learning (ML) model for early sepsis prediction in children.
- To identify key predictors of pediatric sepsis using clinical and cytokine data.
- To enhance the clinical utility of predictive analytics for sepsis.
Main Methods:
- Retrospective cohort study of pediatric patients with infections.
- Utilized clinical data and cytokine profiles (e.g., IL-10, IL-33) for model development.
- Compared various ML models, including logistic regression, and employed SHAP analysis for interpretability.
Main Results:
- Logistic regression model showed superior performance for early sepsis prediction (AUC=0.908).
- Key predictors identified include specific interleukins (IL-10, IL-33) and standard infection markers.
- Shapley Additive Explanations (SHAP) provided model interpretability.
Conclusions:
- An interpretable, high-performance ML model for pediatric sepsis prediction was established.
- The model's integration into an online calculator enables real-time risk assessment.
- This tool can aid early sepsis identification and treatment decisions in clinical settings.
