AUTHOR=Gong Wei , Hu Xiaoxiao , Cui Huimin , Zhao Yuxin , Lin Hong , Sun Peng , Yang Jianjun TITLE=Socioeconomic status and lifestyle as factors of multimorbidity among older adults in China: results from the China Health and Retirement Longitudinal Survey JOURNAL=Frontiers in Public Health VOLUME=Volume 13 - 2025 YEAR=2025 URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2025.1586091 DOI=10.3389/fpubh.2025.1586091 ISSN=2296-2565 ABSTRACT=BackgroundMultimorbidity is increasingly prevalent among older adults and poses significant challenges to public health systems. While previous studies have highlighted the role of individual behaviors, the complex interaction between lifestyle factors and socioeconomic status (SES) in multimorbidity remains unclear.MethodsUsing nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS), we developed predictive models to identify key determinants of multimorbidity among individuals aged ≥60 years. A total of 34,755 participants were included, and 17 features related to demographics, SES, and lifestyle were selected via LASSO regression. Eight machine learning algorithms including logistic regression, decision tree, naive Bayes, neural network, support vector machine, random forest, XGBoost and Bayesian Ridge Regression were applied to build predictive models. Model performance was evaluated using AUC, accuracy, precision, recall, F1-score, RMSE, and decision curve analysis (DCA). SHapley Additive exPlanations (SHAP) were used to interpret model outputs.ResultsXGBoost achieved the best predictive performance (AUC = 0.788 on the test set), outperforming both linear and non-linear models across most evaluation metrics. SHAP analysis revealed that education level, activities of daily living (ADL), work status, self-assessed health status, and per capita income were the top factors associated with of multimorbidity. Subgroup analyses showed variated associations by age and sex, with psychological and geographic factors playing a larger role among those aged ≥80.ConclusionThis study demonstrated the feasibility and interpretability of using machine learning to model complex risk patterns of multimorbidity. Socioeconomic and functional variables were dominant factors associated with multimorbidity, suggesting structural roots of health inequality. These findings offered empirical and theoretical support for early risk stratification and targeted public health interventions aimed at mitigating multimorbidity in aging populations.