Figure 1.

Plots of the most important component loadings from PCA and PLS on 318 anti-CCP-negative RA patients. In these plots, each dot indicates one single patient. Component scores indicate how strongly each component is represented in each patient. For example, in (a), a dot indicates how much the variance in an individual patient is being described by factor 1 on the x-axis (age, gender, and the presence of baseline erosions) in relation to factor 2 on the y-axis (involvement of small joints versus the involvement of large joints or both SJC and CRP). If a concurrence of components was found, clustering of patients would be visible. In the PCA, clinical variables at disease onset were explored. The same applies for the factors in PLS regression. In the PLS regression, the clinical variables at disease onset were explored together with radiologic data on progression of joint destruction during a mean of 5 years of disease. CRP, C-reactive protein; PCA, principal components analysis; PLS, partial least squares regression; RA, rheumatoid arthritis; SJC, swollen joint count.

De Rooy et al. Arthritis Research & Therapy 2011 13:R180   doi:10.1186/ar3505
Download authors' original image