Identifying factors and predicting mental health issues in polypharmacy elderly using machine learning: a study based on the English longitudinal study of aging.

Objectives: Polypharmacy is common among the elderly and is associated with various health issues. Identifying risk factors and developing accurate predictive models for mental health issues in polypharmacy elderly individuals is crucial for early intervention. Methods: Data from the English Longitudinal Study of Aging (ELSA) were used, covering waves 1-9. The study identified factors associated with mental health issues through clinical baseline analysis and univariate logistic regression. Twelve machine learning models were constructed and validated using the mlr3 framework, with KNN emerging as the optimal model based on AUC, DCA, and calibration curves. SHapley Additive exPlanations (SHAP) analysis was used to interpret the KNN model. Results: Baseline analysis and logistic regression identified antidepressant use, age, pain frequency, smoking history, and respiratory improvement medication use as significant factors. The KNN model demonstrated superior performance with an AUC of 0.901 in the training set and 0.827 in the validation set, showing high discrimination ability and robustness. Conclusion: This study identified key factors associated with mental health issues in polypharmacy elderly and validated the KNN model as an effective predictive tool. These findings can inform targeted interventions and improve mental health outcomes in this vulnerable population.