Why the study?
Reliable coronary artery disease risk stratification using routinely available data is critically needed, particularly in clinical settings with limited access to advanced diagnostic resources.
Does a CatBoost machine learning model using minimal clinical data accurately predict angiographically confirmed coronary artery disease in adult patients?
Observational (n=899)
Yes
Does a CatBoost machine learning model using minimal clinical data accurately predict angiographically confirmed coronary artery disease in adult patients?
Answers are grounded in this paper and cite their sources.
A CatBoost machine learning model using 13 routinely available clinical variables provides accurate and well-calibrated prediction of angiographically confirmed coronary artery disease, offering a practical screening tool for resource-limited settings.
Coronary artery disease (CAD) remains a leading cause of morbidity and mortality worldwide. In many clinical settings, particularly those with limited access to advanced diagnostic resources, reliable risk stratification based on routinely available data is critically needed. We analyzed data from 899 patients across four centers (Cleveland, Hungary, Switzerland, and Long Beach) obtained from the UCI Machine Learning Repository. Thirteen routinely collected clinical and demographic variables were used to predict angiographically confirmed CAD (≥50% stenosis). A CatBoost model was developed and evaluated with respect to discrimination, calibration, subgroup performance, and interpretability. Model performance was assessed using accuracy, sensitivity, F1 score, and AUC-ROC. Calibration was examined using calibration curves and the Brier score, while subgroup analyses explored performance differences across sex, age, and center. To assess feasibility in resource-constrained settings, model performance was also evaluated using reduced feature sets. The CatBoost model showed good discriminative performance (accuracy = 0.828, sensitivity = 0.841, F1 = 0.844, AUC-ROC = 0.90) with acceptable overall calibration (Brier score = 0.125). A reduced model using seven variables retained 95.4% of the full model’s F1 performance. Sub-group analyses revealed lower sensitivity in females compared with males and variability across centers, indicating potential fairness concerns. SHAP-based interpretability analysis suggested that chest pain type, ST-segment depression, and ST-segment slope were among the most influential predictors, consistent with established clinical understanding of CAD. A CatBoost model based on a small set of routinely available clinical variables can support accurate and reasonably well-calibrated CAD risk stratification. However, observed subgroup disparities highlight the importance of fairness evaluation before clinical deployment. The proposed minimal-data and explainable framework may offer a practical approach for CAD screening, particularly in resource-limited settings.
No takes yet. Share an insight, caveat, or question.
Alisoufi et al. (2026) conducted an observational in Coronary Artery Disease (n=899). CatBoost machine learning model vs. Other machine learning models (XGBoost, Random Forest, Logistic Regression, Decision Tree) was evaluated on F1 score for predicting angiographically confirmed CAD (≥50% stenosis) (95% CI 0.817-0.867).
Synapse has enriched 10 related papers. Consider them for comparative context: