Explainable Machine Learning to Detect Osteoarthritis Comorbidity in Adults With Diabetes: A Nationally Representative NHANES Analysis

1

This cross-sectional analysis used seven NHANES cycles from 2005 to 2018 to study osteoarthritis comorbidity among adults with diabetes.
2

The final analytic sample included 6,716 participants with diabetes after exclusions for missing disease, dietary, education, or covariate information.
3

Dietary nutrient intake was assessed using two 24-hour recalls, with values averaged for participants who completed both recalls.
4

The study combined weighted quantile sum regression and explainable machine learning to examine nutrient-mixture associations and classify osteoarthritis comorbidity.
5

The analysis compared demographic-plus-dietary and dietary-only models and used SHAP and LIME to assess individual nutrient contributions to classification.

1