Related Experiment Video
Updated: Jun 9, 2025
08:10
Estimating Bilateral Atrial Function by Cardiovascular Magnetic Resonance Feature Tracking in Patients with Paroxysmal Atrial Fibrillation
Published on: July 20, 2022
1.6K
Interpretable machine learning model for new-onset atrial fibrillation prediction in critically ill patients: a
Chengjian Guan1, Angwei Gong1, Yan Zhao1
1Department of Cardiology, The Second Hospital of Hebei Medical University, Shijiazhuang, 050000, People's Republic of China.
Critical Care (London, England)
|October 30, 2024
Summary
We developed a machine learning model to predict new-onset atrial fibrillation (NOAF) in critically ill patients. This tool helps identify high-risk individuals for early intervention and improved outcomes.
Area of Science:
- Critical Care Medicine
- Biomedical Informatics
- Machine Learning in Healthcare
Background:
- New-onset atrial fibrillation (NOAF) is a common complication in intensive care units (ICUs), associated with adverse patient outcomes.
- Early identification of critically ill patients at high risk for NOAF is essential for timely intervention.
- Existing prediction methods for NOAF in this population require enhancement.
Purpose of the Study:
- To develop and validate a machine learning (ML) model for predicting the risk of NOAF in critically ill patients.
- To identify key predictors of NOAF in this patient cohort.
- To create a clinically applicable tool for risk stratification.
Main Methods:
- Utilized two large, non-overlapping datasets from the Medical Information Mart for Intensive Care (MIMIC) database (MIMIC-IV for training, MIMIC-III for external validation).
- Employed LASSO regression for feature selection and evaluated eight ML algorithms, with XGBoost selected as the optimal model.
- Assessed model performance using identification, calibration, and clinical utility metrics, incorporating SHapley Additive exPlanations (SHAP) for interpretability.
Main Results:
- The final ML model incorporated 23 variables and demonstrated strong predictive performance, with an AUC of 0.891 in internal validation and 0.769 in external validation.
- Key predictors identified include age, mechanical ventilation, urine output, sepsis, blood urea nitrogen, oxygen saturation, continuous renal replacement therapy, and weight.
- A user-friendly interface was developed, defining a risk probability > 0.6 as high risk.
Conclusions:
- A validated ML model can reliably predict the risk of NOAF in critically ill patients not undergoing cardiac surgery.
- The use of SHAP values enhances model interpretability, aiding clinicians in understanding NOAF risk factors and facilitating preventative strategies.
- This predictive tool has the potential to improve clinical decision-making and patient outcomes in the ICU setting.
