Baseline characteristics of the training and test set populations (case and control)1.

nd, not determined; ACPA, anti-citrullinated protein antibodies; RF, rheumatoid factor; DAS28, 28 joint count disease activity score; HAQ, health assessment questionnaire score; CRP, C-reactive protein.1Differences between training set and test set within both RA and control groups were not statistically significant (Mann-Whitney for continuous values and Chi Square for categorical values; p2Difference in the median age value between groups is statistically significant between RA and control groups of the training set (p = 0.0023) and between RA and control groups of the test set (p = 0.0059).3Difference in the gender distribution between groups is statistically significant between RA and control groups of the training set (p0.05) (Chi Square test).4Data missing for 18 patients in the training set and 9 patients in test set, percentage refers to proportion of patients tested.