Inflammatory biomarkers partially explain the association between dietary index for gut microbiota and all-cause and cardiovascular mortality in patients with cardiovascular disease: A prospective cohort study and machine learning model

There is growing interest in the link between diet, the gut microbiota, and cardiovascular disease (CVD); the impact of microbiota-targeted diets on long-term survival remains poorly understood. This study investigates the association between the dietary index for gut microbiota (DI-GM) and all-cause and cardiovascular mortality in CVD patients and explores the potential explanatory role of inflammatory biomarkers. We analyzed data from 1558 CVD patients in the National Health and Nutrition Examination Survey (NHANES). Cox proportional hazards models and restricted cubic splines (RCS) were employed to examine the association between DI-GM and mortality risk, and to assess potential non-linearity. The proportion statistically accounted for by inflammatory biomarkers was estimated using counterfactual-based marginal structural models. Survival machine-learning models assessed the prognostic value of the DI-GM, employing Shapley Additive Explanations (SHAP) for interpretability. In fully adjusted Cox models, the highest DI-GM category (≥6) was significantly associated with lower risks of all-cause (HR = 0.70; 95% CI, 0.56–0.87; P < 0.001) and cardiovascular mortality (HR = 0.63; 95% CI, 0.54–0.79; P < 0.001). RCS analyses confirmed a linear inverse relationship for the DI-GM, whereas the UGM score exhibited significant nonlinear dose-response effects. Analyses indicated that inflammatory biomarkers statistically accounted for the associations between DI-GM and mortality, with explained proportions from 10.1% to 20.7%. The Random Survival Forest (RSF) model demonstrated the best predictive performance and provided significant incremental prognostic value. SHAP analysis identified fermented dairy, coffee, fat, whole grains, and refined grains as the most influential risk predictors. Higher DI-GM was associated with lower risks of all-cause and cardiovascular mortality in CVD patients, and inflammatory biomarkers partially explained this association. Machine-learning models improved component-level stratification within the DI-GM framework, which may help enable precision risk stratification and personalized nutritional interventions in secondary cardiovascular prevention. Graphical abstract • Higher DI-GM scores were associated with lower risks of all-cause and cardiovascular mortality. • Systemic inflammation partially explains this protective diet-survival link. • Integrating DI-GM into survival machine learning improves prognostic accuracy. • SHAP identifies fermented dairy and coffee as key protective functional foods.