Unraveling the mechanisms of health management needs among rural elderly in underdeveloped Chinese regions: a machine learning approach to predictive model building and factor analysis

Global population aging is accelerating, with China among the countries experiencing the fastest and largest aging populations. This study aims to analyze the current health management(HMN) needs of rural elderly in underdeveloped regions of China. Machine learning algorithms were employed to construct predictive models and identify key influencing factors, providing empirical evidence for the development of targeted health management strategies. From August 2023 to January 2024, a convenience sample of 641 rural community elderly aged ≥ 60 years across four prefecture-level cities in Guangxi was surveyed using questionnaires. Predictor variables were chosen using LASSO regression and Logistic regression. The predictive efficacy of three machine learning models - Logistic regression, Random Forest, and XGBoost - was methodically evaluated. Variable contributions were assessed using SHAP values, and model validity and practical applicability were validated through the nomogram, ROC curve,calibration curves, and decision curve analysis. Health management needs among rural elderly in China’s underdeveloped regions were relatively low (43.84%). SHAP interpretability analysis identified five key factors influencing health management needs. By conducting SHAP explainability analysis, five crucial factors affecting the healthcare requirements of the elderly were pinpointed. XGB exhibited superior predictive accuracy in both the training and validation datasets, achieving AUC values of 0.783 and 0.723, respectively.Independent samples t-tests revealed six critical individual factors affecting health management needs. The study identified three primary health management needs: regular health monitoring and screening, enhanced health management education, and targeted chronic disease management services. The XGB prediction model constructed in this study effectively identifies health management needs among rural elderly in underdeveloped regions of China. It is recommended that relevant authorities optimize the allocation of rural community health service resources and implement targeted health interventions based on the high-demand population characteristics identified by the model to promote active aging.