Abstract
INTRODUCTION: Anxiety and depression are common among hypertensive patients and can lead to significant health complications. This study aimed to use Extreme Gradient Boosting (XGB) machine learning (ML) technique to select associated factors of anxiety and depression symptoms among people with hypertension in rural areas. METHODOLOGY: A cross-sectional study was conducted using a multistage cluster random sampling. The anxiety and depression symptoms were evaluated using the Generalized Anxiety Disorder-7 (GAD-7) and Patient Health Questionnaire-9 (PHQ-9) scales, respectively. A chi-square test was performed to assess prevalence. XGB model was employed to predict the presence of anxiety and depression symptoms using 13 variables, and the model's performance was compared with that of the traditional logistic regression (LR) model. Influential variables were explained and ranked using SHapley Additive exPlanations (SHAP) technique. RESULTS: Among the 496 rural hypertensive adults, approximately 5.9% and 6.4% experienced the presence of anxiety and depression symptoms, respectively. Anxiety and depression symptoms were more prevalent among higher educated patients (14.0%) and who used tobacco (12.4%), respectively. The XGB model demonstrated improved predictive performance (for anxiety, ROC for XGB: 93.1%; for depression, ROC for XGB: 90.7%) compared to the LR model (for anxiety, ROC for LR: 83.8%; for depression, ROC for XGB: 79.7%) in predicting both outcomes. Marital status, body mass index (BMI), cardiovascular disease (CVD), educational status, family history of hypertension and employment were the influential factors in predicting the presence of anxiety symptoms. Similarly, chewing tobacco, family history of hypertension, marital status, CVD, sex, and educational status are important factors in predicting the presence of anxiety. CONCLUSION: In Bangladesh, around 6% rural individuals with hypertension experienced the presence of anxiety and depression symptoms. Educational status, marital status, CVD and family history of hypertension were key factors linked to both outcomes. Future research is needed to validate these findings.