The Association of Elevated Depression Levels and Life’s Essential 8 on Cardiovascular Health With Predicted Machine Learning Models and Interpretations: Evidence From NHANES 2007–2018

Abstract

Background and Objective: The association between depression severity and cardiovascular health (CVH) represented by Life’s Essential 8 (LE8) was analyzed, with a novel focus on ranked levels and different ages. Machine learning (ML) algorithms were also selected aimed at providing predictions to suggest practical recommendations for public awareness and clinical treatment. Methods: We included 21,279 eligible participants from the National Health and Nutrition Examination Survey (NHANES) 2007–2018. Weighted ordinal logistic regression (LR) was utilized with further sensitivity and dose–response analysis, and ML algorithms were analyzed with SHapley Additive exPlanations (SHAP) applied to make interpretable results and visualization. Results: Our studies demonstrated an inverse relationship between LE8 and elevated depressive levels, with robustness confirmed through subgroup and interaction analysis. Age-specific findings revealed middle-aged and older adults (aged 40–60 and over 60) which showed higher depresion severity, highlighting the need for greater awareness and targeted interventions. Eight ML algorithms were selected to provide predictive results, and further SHAP would become ideal supplement to increase model interpretability. Conclusions: Our studies demonstrated a negative association between LE8 and elevated depressive levels and provided a suite of ML predictive models, which would generate recommendations toward clinical implications and subjective interventions.