Research on coronary heart disease risk factors based on Bayesian network: a cross-sectional survey of 129,155 individuals

Introduction Identifying coronary heart disease (CHD) risk factors is crucial for prevention and control. While existing studies using logistic regression identify individual factors, they inadequately quantify complex interactions. This study constructs a Bayesian network (BN) model to analyze intricate relationships among CHD risk factors and quantify multifactorial interactions through probabilistic inference.Theory The core framework is BN. Based on probabilistic graphical models, BN enables uncertainty reasoning, overcoming limitations of traditional linear models and better reflecting the complex biology of disease.Method The study enrolled 129,115 participants from 9 cities and 13 counties in Shanxi Province, China. Data were collected via questionnaires, physical examinations, and laboratory tests. Fourteen variables were selected for the model using chi-square tests and logistic regression. The BN structure was learned using the Max-Min Hill-Climbing (MMHC) algorithm, with parameters estimated via maximum likelihood estimation.Result s: The BN identified age, dyslipidemia, diabetes, and family history of CHD as direct risk factors for CHD. Snoring, hypertension, family history of hypertension, and abdominal obesity were indirect risk factors. Probabilistic inference showed the baseline CHD prevalence of 1.68% increased to 2.23% with dyslipidemia, rose further to 3.72% when combined with age (60–75 years), and reached 5.28% when diabetes was also present.Discussion The findings are broadly consistent with prior epidemiological evidence while quantifying the synergistic effects of key risk factors through probabilistic inference. Future research should incorporate additional variables and validate the model in diverse populations.