Cognitive function and dietary index for gut microbiota in the aging population: association and machine learning screening models

Background

Accelerated global population aging has heightened cognitive impairment as a critical public health challenge. Emerging evidence implicates gut microbiota in cognitive regulation via the gut-brain axis, with diet being a key modifiable factor. The Dietary Index for Gut Microbiota (DI-GM) quantifies diet-microbiota interactions, but its link to cognition remains unexplored. This study investigated the DI-GM-cognition relationship in older adults and evaluated its predictive value for cognitive impairment.

Methods

Data from 2168 older participants in the 2011–2014 NHANES were analyzed. The association between DI-GM and cognitive function was assessed using linear regression with smoothed curve fitting. The robustness and heterogeneity of the results were evaluated through sensitivity and subgroup analyses. A cognitive impairment screening model was developed using three machine learning algorithms (XGBoost, AdaBoost, Random Forest), and key associated features in the optimal model were identified via SHAP analysis.

Results

After full covariate adjustment, higher DI-GM scores significantly correlated with better cognitive function (β = 1.25, 95 % CI: 0.62–1.88, p = 0.0001) and reduced cognitive impairment probability (OR = 0.91, 95 % CI: 0.84–0.99, p = 0.023). Subgroup analyses confirmed consistency across populations, with marital status moderating effects. XGBoost achieved optimal prediction (AUC = 0.87); SHAP analysis identified age, diabetes, and DI-GM as key associated features.

Conclusion

DI-GM positively associates with cognitive function and negatively with impairment probability, suggesting diet targeting gut microbiota may support cognitive protection. Key associated features (age, diabetes, DI-GM) highlight the importance of dietary optimization and chronic disease management for cognitive health in aging populations.