Development and validation of an interpretable machine learning model for predicting atrial fibrillation risk in middle-aged and older patients with coronary heart disease - Report - MDSpire
Advertisement
Development and validation of an interpretable machine learning model for predicting atrial fibrillation risk in middle-aged and older patients with coronary heart disease
Clinical Report: Explainable Machine Learning for Atrial Fibrillation Risk Forecasting
Overview
This study developed an interpretable machine learning model to predict atrial fibrillation (AF) risk in middle-aged and elderly patients with coronary heart disease (CHD). The XGBoost model was evaluated for its performance in predicting AF risk.
Background
Coronary heart disease (CHD) and atrial fibrillation (AF) often coexist, complicating patient management and worsening clinical outcomes. Current risk stratification tools inadequately address the complex interactions of risk factors in this population. There is a need for predictive models that can leverage electronic medical records to identify high-risk individuals.
Data Highlights
The study analyzed a cohort of 47,617 hospitalized CHD patients. Key findings include:
Key Findings
LASSO regression identified 16 predictors for AF risk, with pulse rate and total cholesterol among the top contributors.
The XGBoost model achieved an AUC of 0.867 (95% CI: 0.862–0.872) in the training set.
In the validation set, the model maintained an AUC of 0.813 (95% CI: 0.802–0.823).
Restricted cubic spline analysis indicated nonlinear dose–response relationships for several continuous variables.
SHAP and LIME methods were utilized for model interpretability, providing insights into individual feature contributions.
Clinical Implications
The developed XGBoost model provides a data-driven approach for AF risk assessment in CHD patients.
Conclusion
This study presents a machine learning framework for predicting AF risk in CHD patients.
