Related Experiment Video
Updated: Aug 8, 2026
06:13
The Intra-Aortic Balloon Pump
Published on: February 5, 2021
Interpretable Machine Learning Models for Mortality Prediction in Critically Ill Patients with Intra-Aortic Balloon
Kai Zhang1, Jie Min2, Lei Zhong2
1Department of Emergency, Huzhou Central Hospital, Affiliated Central Hospital of Huzhou University, Huzhou, China; Huzhou Central Hospital, Fifth School of Clinical Medicine of Zhejiang Chinese Medical University, Huzhou, China.
Heart, Lung & Circulation
|August 6, 2026
Summary
A new machine learning model accurately predicts mortality risk in intensive care unit (ICU) patients receiving intra-aortic balloon pump (IABP) therapy. This tool aids clinicians in better patient risk stratification and decision-making.
Area of Science:
- Critical Care Medicine
- Biomedical Informatics
- Machine Learning in Healthcare
Background:
- Intra-aortic balloon pump (IABP) therapy is used for critically ill patients.
- Accurate mortality risk prediction is crucial for managing these patients.
Purpose of the Study:
- Develop and validate predictive models for mortality risk in ICU patients undergoing IABP therapy.
- Identify key predictors of mortality in this patient population.
Main Methods:
- Retrospective analysis of 764 ICU patients from the MIMIC-IV database.
- Utilized LASSO regression for variable selection and constructed six machine learning models.
- Evaluated models using ROC curves, calibration curves, and decision curve analysis.
Main Results:
- Identified nine key mortality predictors including lactate, norepinephrine, and renal replacement therapy.
- The Random Forest (RF) model showed superior performance with an AUC of 0.965 (training) and 0.900 (validation).
- The RF model demonstrated clinical utility for predicting ICU mortality in IABP patients.
Conclusions:
- An interpretable machine learning model (RF) effectively predicts mortality in critically ill IABP patients.
- This model aids in risk stratification, supporting informed clinical decision-making.
- The developed model can be implemented via a web-based calculator for practical use.