Summary
Many heart disease screening models seem very accurate, but this paper shows that the high scores come from including data that reveal the diagnosis (target leakage), not from better machine learning methods. The authors tested various models on a large health survey and found removing these revealing features reduced accuracy for all models equally. A transparent, interpretable model performed just as well but was much faster and helped fix fairness issues in predictions. This means choosing the right input data and careful evaluation matter more than the model type for reliable heart disease screening.
What this means in practice
- •For health data analysts: Use transparent models and audit feature sets to ensure cardiovascular screening models do not inflate accuracy with leaked information.
- •For clinical decision support teams: Deploy efficient and interpretable screening models that maintain accuracy while reducing fairness gaps in predicting heart disease risks.
Authors
Raad Bin Tareaf, Murad Al-Rajab, Samia Loucif, Samer Ellaham, Cedric Schmitz
Abstract
Cardiovascular screening models trained on national health surveys routinely report areas under the receiver operating characteristic curve (AUROC) near 0.89. We asked whether that accuracy reflects learning or target leakage, whether tabular foundation models change the answer, and whether the properties deployment requires survive joint examination. We benchmarked ten classifiers spanning linear, tree-ensemble, neural, glass-box, and tabular foundation classes for prevalent myocardial infarction in 442,067 respondents of the 2022 Behavioral Risk Factor Surveillance System across five feature tiers of decreasing leakage risk. Each was audited for discrimination, calibration, fairness at an explicit screening threshold, conformal coverage, explanation faithfulness, and inference cost, then applied -- models and thresholds frozen -- to 430,755 respondents of 2023. Removing two post-diagnostic features cost every model 0.049-0.051 AUROC, collapsing the field into a 0.0045-wide band. The glass-box explainable boosting machine was non-inferior to every alternative within a pre-specified 0.005 margin while scoring the cohort roughly 104 times faster than the strongest foundation model. One threshold detected 75.4% of women's infarctions against 89.0% of men's; editing the model's shape functions reduced the gap to 0.010. Marginal conformal prediction gave 0.86 coverage to men and 0.82 to adults over 60; Mondrian calibration repaired every stratum. Frozen models transported within 0.002 AUROC. Reported headroom in this literature is a property of the feature set, not the learner. Transparency cost nothing measurable and made fairness repair and uncertainty conditioning directly auditable. Evaluation practice, not model capacity, is the binding constraint.