Key result
XGBoost model outperforms logistic regression for AMI prediction with an AUC of 0.89.
Why the study?
Although IHD mortality has declined in China, incidence has not decreased in parallel, and the global risk-factor profile is shifting toward metabolic drivers.
Does an XGBoost machine learning model improve AMI risk stratification compared to multiple logistic regression in adults with atherosclerotic coronary heart disease?
Cohort (n=5,597)
Yes
Does an XGBoost machine learning model improve AMI risk stratification compared to multiple logistic regression in adults with atherosclerotic coronary heart disease?
Machine learning using XGBoost significantly improves acute myocardial infarction risk stratification compared to traditional logistic regression, highlighting the predictive importance of CRP, BMI, and age.
Ischemic heart disease (IHD) remains the leading global cause of mortality. Although IHD mortality has declined in China over recent decades, incidence has not declined in parallel. At the same time, the global IHD risk-factor profile is shifting from traditional behavioral and environmental exposures toward metabolic drivers. We analyzed IHD burden using Global Burden of Disease (GBD) 2021 estimates, comparing mortality, incidence, disability-adjusted life years (DALYs), and attributable risk factors across China, the United States, and global averages. We then analyzed a large national health examination cohort of 186,838 adults; among 5,597 participants with atherosclerotic coronary heart disease and complete AMI information, AMI status was modeled using multiple logistic regression and Extreme Gradient Boosting (XGBoost). China’s age-standardized IHD mortality decreased by 32.1% between 2000 and 2021, while the age-standardized incidence rate increased by 37.5% from 1990 to 2020. The attributable burden shifted toward metabolic risks, including high systolic blood pressure, high LDL cholesterol, high fasting plasma glucose, and high BMI. In the cohort analysis, hypertension, older age, and higher C-reactive protein (CRP) levels were independently associated with AMI in multiple logistic regression. XGBoost showed excellent discrimination for AMI prediction (AUC = 0.89), outperforming multiple logistic regression (AUC = 0.776). Gain-based feature importance identified CRP, BMI, and age as the leading predictors. Declining IHD mortality, rising incidence, and increasing metabolic attribution highlight persistent gaps in primary prevention. Machine learning, particularly XGBoost, may improve AMI risk stratification in large health examination datasets, but model-derived importance should be interpreted as predictive rather than causal and requires further external validation.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) conducted a cohort in Atherosclerotic coronary heart disease (n=5,597). XGBoost machine learning model vs. Multiple logistic regression was evaluated on Area under the ROC curve (AUC) for acute myocardial infarction prediction. The XGBoost machine learning model achieved excellent discrimination for predicting acute myocardial infarction with an AUC of 0.89, outperforming multiple logistic regression (AUC 0.776).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:
More from Journal of Cardiothoracic Surgery