Abstract
BACKGROUND: Sjögren's Syndrome (SS) and Rheumatoid Arthritis (RA) are autoimmune conditions that often coexist in female patients. Biomarkers such as antinuclear antibodies (ANA) and rheumatoid factor (RF) are used for diagnosis, but their predictive power for treatment outcomes remains unclear. This study aims to investigate the correlation between age, ANA, RF, and treatment response in female patients with both SS and RA. OBJECTIVE: To evaluate the relationships between age, ANA, RF levels, RA (disease present), and treatment response using Pearson correlation analysis and a neural network model, to predict treatment outcomes in patients with coexisting SS and RA. METHODS: A cohort of 56 female patients aged 30-73 was analyzed. Descriptive statistics provided an overview of key variables, followed by Pearson correlation analysis to assess relationships between age, ANA, RF, RA, and treatment response. A neural network model was developed to predict treatment response based on age, ANA, and RF levels, using a training-to-testing split of 81.3 % and 18.8 %, respectively. RESULTS: The Pearson correlation analysis revealed a significant positive correlation between age and ANA levels (r = .541, p = 0.031), though no significant correlations were found between age, RF, RA, and treatment response. The neural network model achieved an accuracy of 92.3 % during training and 100 % accuracy during testing for most treatment categories. However, the model struggled to accurately distinguish between certain classes, particularly treatment categories 1 and 3. CONCLUSION: Age showed a significant correlation with ANA levels, indicating that older patients may have elevated ANA. The neural network model demonstrated strong predictive power for treatment response, although further refinement is needed to improve its ability to distinguish between all response categories. These findings suggest that machine learning models could enhance personalized treatment strategies for patients with SS and RA, but additional validation with larger datasets is required.