Related Experiment Video
Updated: Feb 28, 2026
Uncontrolled Hemorrhagic Shock Modeled via Liver Laceration in Mice with Real Time Hemodynamic Monitoring
Published on: May 21, 2017
An interpretable machine learning model with SHAP explanations predicts spontaneous bleeding in pediatric acute liver
Qiang Xiong1, Ruijue Wang1, Chenyu Yang1
1Department of Hepatobiliary Surgery, National Clinical Research Center for Child Health and Disorders, Ministry of Education Key Laboratory of Child Development and Disorders, Chongqing Key Laboratory of Pediatrics, Children's Hospital of Chongqing Medical University, Chongqing, China.
A new machine learning model accurately predicts spontaneous bleeding in pediatric acute liver failure (PALF) patients. This tool aids clinicians in identifying high-risk children and improving outcomes for PALF bleeding risk.
Area of Science:
- Pediatric Hepatology
- Medical Informatics
- Machine Learning in Medicine
Background:
- Pediatric acute liver failure (PALF) presents a significant risk of spontaneous bleeding, contributing to increased mortality.
- Current bleeding risk prediction methods for PALF are insufficient, necessitating advanced predictive tools.
- Machine learning (ML) offers a promising avenue for developing accurate and interpretable models for PALF complications.
Purpose of the Study:
- To develop and validate a machine learning model for predicting spontaneous bleeding in pediatric patients diagnosed with PALF.
- To enhance the interpretability of the predictive model using SHapley Additive exPlanations (SHAP) for clinical utility.
Main Methods:
- A retrospective cohort study utilized data from 501 PALF patients for training and 153 for external validation.
- Feature selection was performed using Boruta and LASSO regression, with Gradient Boosting Machine (GBM) selected from ten ML algorithms.
- Model performance was evaluated using AUC, accuracy, recall, specificity, precision, F1 score, Brier score, calibration curves, and decision curve analysis (DCA), with SHAP values for interpretability.
Main Results:
- The GBM model demonstrated strong predictive performance with an AUC of 0.858 (internal) and 0.839 (external validation).
- Key predictors identified included platelet count, infection, multiple organ dysfunction syndrome (MODS), hepatorenal syndrome (HRS), D-dimer, total protein, and lactic acid.
- SHAP analysis revealed infection, MODS, and HRS as risk factors, while platelet count, total protein, and fibrinogen showed protective effects.
Conclusions:
- The developed ML model provides robust and interpretable prediction of spontaneous bleeding in PALF patients.
- This tool can assist clinicians in early identification of high-risk individuals, guiding timely interventions.
- Further validation with diverse datasets and exploration of bleeding severity prediction are recommended.
