JMIR Publications

ABSTRACT

Background:

Research has revealed potential links between specific dietary patterns and accelerated aging. However, most studies focus only on singular diets or lack ethnic diversity.

Objective:

This study aims to investigate the associations between five dietary habits and the risk of accelerated aging, and develop an interpretable machine learning model for accelerated aging prediction.

Methods:

Our study explored the associations between diet indices and the risk of accelerated aging using data from the National Health and Nutrition Examination Survey. A weighted linear regression analysis was used to determine if accelerated aging was linked to dietary habits, and the covariates were gradually adjusted to ensure that the association was stable. Nonlinear correlations were explored using Restricted Cubic Splines. In addition, multiple machine learning algorithms were used to build predictive models of accelerated aging risk.

Results:

Except for the dietary inflammation index (DII) [β (95% CI): 0.275(0.325,0.226)], the other four dietary indicators (AHEI, aMED, HEI-2020, and DASH) were negatively associated with the risk of accelerated aging. A nonlinear relationship was found between DII and accelerated aging calculated by KDM (Poverall < 0.001). Ten machine learning algorithms were used to establish risk prediction models and SVM model has the best performance (AUC=0.743). An online prediction platform was available at http://mitusml.com:9538/.

Conclusions:

Significant associations between accelerate aging and dietary indices were observed. DII high dietary compliance has a promoting effect on accelerated aging, while AHEI, aMED, HEI-2020 and DASH high dietary compliance have different degrees of protective effect on accelerated aging.