Key result
XGBoost yields minimal CVD classification gains over standard logistic regression across 46 countries.
Why the study?
Machine-learning models for cardiovascular disease are often evaluated with random participant splits, which may overstate performance when countries differ in prevalence, risk-factor distributions, healthcare access, and survey implementation.
Do complex machine-learning classifiers (such as XGBoost and random forest) improve classification discrimination, calibration, and geographic and temporal transportability for prevalent self-reported cardiovascular disease compared to standard logistic regression across 46 countries?
Cross-Sectional (n=187,791)
Yes
Do complex machine-learning classifiers (such as XGBoost and random forest) improve classification discrimination, calibration, and geographic and temporal transportability for prevalent self-reported cardiovascular disease compared to standard logistic regression across 46 countries?
Complex machine-learning models like XGBoost yield negligible pooled discrimination gains over standard logistic regression for global cardiovascular disease classification while exhibiting substantial geographic and temporal instability, underscoring the need to evaluate calibration and transportability rather than pooled ROC AUC alone.
Answers are grounded in this paper and cite their sources.
Negligible XGBoost gains over logistic regression should not change practice; leaves open whether complex models improve transportability or calibration in CVD classification.
Background Machine-learning models for cardiovascular disease (CVD) are often evaluated with random participant splits, which may overstate performance when countries differ in prevalence, risk-factor distributions, healthcare access, and survey implementation. We benchmarked statistical and machine-learning classifiers for prevalent self-reported CVD with explicit assessment of class imbalance, calibration, explainability, and geographic and temporal transportability. Methods We analysed harmonised WHO STEPwise approach to noncommunicable disease risk factor surveillance (STEPS) surveys conducted from 2013 to 2023. The cohort included 187,791 adults, including 16,305 CVD cases, from 46 countries. Logistic regression, elastic-net logistic regression, random forest, and XGBoost were evaluated using five-fold country-grouped internal-external cross-validation with survey, country-year, and class weighting. We assessed ROC AUC, precision-recall AUC (PR AUC), Brier score, calibration, threshold-dependent performance, SHAP importance, and earlier-to-later transportability. Results Weighted CVD prevalence was 7.78%. XGBoost achieved ROC AUC 0.671, PR AUC 0.162, and Brier score 0.0692; logistic regression achieved 0.669, 0.154, and 0.0694. Country-level performance varied substantially. Removing hypertension and diabetes reduced ROC AUC by 0.038–0.040. Temporal logistic-regression ROC AUC was 0.634 in Mongolia, 0.579 in Uganda, and 0.670 in Viet Nam. Conclusions Complex models produced only small pooled gains over logistic regression, whereas geographic and temporal heterogeneity was substantial. Responsible public-health AI benchmarking should prioritise minority-class performance, calibration, robustness, explainability, and transportability rather than pooled ROC AUC alone.
No takes yet. Share an insight, caveat, or question.
Ng’ambi et al. (2026) conducted a cross-sectional in Cardiovascular disease (n=187,791). XGBoost machine learning classifier vs. Logistic regression classifier was evaluated on Classification of prevalent self-reported cardiovascular disease (survey-weighted ROC AUC). XGBoost achieved only small pooled gains over logistic regression for classifying prevalent CVD across 46 countries.
Synapse has enriched 10 related papers. Consider them for comparative context: