# Causal Effects of Lifestyle and Dietary Factors on Rheumatoid Arthritis: An Integrated Analysis Combining Mendelian Randomization, Machine Learning, and Evaluation of Burden Dynamics and Health Inequality

## TL;DR

This study finds that lifestyle and dietary factors like obesity and smoking increase rheumatoid arthritis risk, while others like cheese intake may protect against it, and machine learning helps predict RA risk effectively.

## Contribution

The study integrates Mendelian randomization, machine learning, and global burden analysis to identify causal lifestyle and dietary links to rheumatoid arthritis and assess health inequalities.

## Key findings

- Obesity, smoking, and poultry intake increase RA risk, while pork and cheese intake are protective.
- Random forest machine learning achieved high RA risk prediction accuracy with age as the top predictor.
- Global RA burden is projected to rise through 2050 due to aging populations and persistent health disparities.

## Abstract

Rheumatoid arthritis (RA) is a chronic autoimmune disease contributing to global morbidity and disability. Despite its growing burden, causal links between lifestyle, dietary factors, and RA remain unclear. This study investigates causal links between lifestyle, dietary factors, and RA using Mendelian randomization (MR) and machine learning (ML). Two‐sample MR analyzed 42 lifestyle and dietary exposures, while an RA risk prediction model was developed using nationally representative data and nine ML algorithms. Global Burden of Disease data assessed health inequalities and RA burden trends through frontier, decomposition, and Bayesian Age‐Period‐Cohort analyses. MR identified obesity, current smoking, sleeplessness, poultry intake, and salt added to food as RA risk factors, while never smoking, pork consumption, and cheese intake were protective. Random forest showed superior predictive performance, with age as the most influential predictor; seven features exhibited nonlinear RA risk associations. RA burden revealed gender and regional disparities, with frontier analysis indicating potential for burden reduction in multiple countries. Between 1990 and 2021, global RA burden rose due to population growth and aging, with projections suggesting continued increases through 2050. These findings highlight the importance of targeted lifestyle and dietary interventions to reduce RA burden and address health inequities in high‐risk populations.

Using two‐sample Mendelian randomization, higher BMI, obesity class 2, current smoking, sleeplessness, poultry intake, and salt added to food were causally associated with increased rheumatoid arthritis (RA) risk, while never smoking, pork intake, and cheese intake showed protective associations. A machine‐learning framework identified seven key predictors and achieved robust discrimination (random forest AUROC = 0.908) with good calibration and net benefit. Global burden analyses revealed increasing RA prevalence and DALYs from 1990 to 2021 with persistent health inequalities, and projections suggest continued growth through 2050.