Speakers - DEWC2025

Abstract

Background:

The gut microbiota plays a crucial role in modulating host metabolism and cardiovascular risk, especially in individuals with metabolic syndrome (MetS). The Dietary Index for Gut Microbiota (DI-GM) is a novel scoring system designed to quantify the functional impact of diet on microbial health. However, its prognostic utility in relation to long-term mortality risk remains unclear.

Methods:

We analyzed data from the U.S. National Health and Nutrition Examination Survey (NHANES, 2003–2018), involving 7,939 adults diagnosed with MetS. The DI-GM score was constructed based on 14 food components with experimentally validated effects on gut microbiota. Cox proportional hazards models were applied to assess the association between baseline DI-GM and all-cause and cardiovascular mortality across three progressively adjusted models. Restricted cubic spline (RCS) regression explored dose–response relationships, and subgroup analyses evaluated interaction effects. Additionally, eight machine learning models were developed to predict mortality risk, and SHapley Additive exPlanations (SHAP) were used to interpret the contributions of individual dietary factors.

Results:

Higher DI-GM scores were significantly associated with reduced all-cause (HR per unit increase: 0.944; 95% CI: 0.897–0.993; P = 0.027) and cardiovascular mortality (HR: 0.907; 95% CI: 0.837–0.984; P = 0.018) in fully adjusted models. A significant linear inverse relationship was observed between DI-GM and mortality outcomes without evidence of nonlinearity. Subgroup analyses indicated stronger protective effects among older adults, women, those with hypertension or elevated BMI, and individuals without diabetes. The SHAP framework identified fiber and coffee as the most protective dietary components, while the effects of cranberry were context-dependent. Among predictive models, surv.glmnet performed best for all-cause mortality (C-index: 0.812), while the Cox model excelled in predicting cardiovascular mortality (C-index: 0.855).

Conclusion:

DI-GM is a microbiota-informed, functionally grounded dietary index that independently predicts mortality risk in individuals with MetS. Its integration with explainable machine learning enhances transparency in dietary risk assessment and offers a promising platform for personalized nutrition and early intervention in cardiometabolic health management.