Abstract
INTRODUCTION: Epigenetic biomarkers are molecular indicators of epigenetic changes, and some studies have suggested that these biomarkers have predictive power for disease risk. This study aims to analyze the relationship between 30 epigenetic biomarkers and the risk of diabetes and cancer using machine learning modeling. METHODS: The data for this study were sourced from the NHANES database, which includes DNA methylation arrays and epigenetic biomarker datasets. Nine machine learning algorithms were used to build models: AdaBoost, GBM, KNN, lightGBM, MLP, RF, SVM, XGBoost, and logistics. Model stability was evaluated using metrics such as Accuracy, MCC, and Sensitivity. The performance and decision-making ability of the models were displayed using ROC curves and DCA curves, while SHAP values were used to visualize the importance of each epigenetic biomarker. RESULTS: Epigenetic age acceleration was strongly associated with cancer risk but had a weaker relationship with diabetes. In the diabetes model, the top three contributing features were logA1Mort, family income-to-poverty ratio, and marital status. In the cancer model, the top three contributing features were gender, non-Hispanic White ethnicity, and PACKYRSMort. CONCLUSION: Our study identified the relationship between epigenetic biomarkers and the risk of diabetes and cancer, and used machine learning techniques to analyze the contributions of various epigenetic biomarkers to disease risk.