Fig. 1

Immunological phenotyping principal component analysis (PCA) linked to left ventricle ejection fraction (LVEF) tertiles categorization. a PCA of the 19 cytokines variables obtained from blood sampled of 34 COVID-19 ICU patients. Displayed as graph of the variables, correlation circle is represented around the first plan PCA: the first axis (Dim1) being orthogonal to the second axis (Dim2). Each vector (arrow) of these 19 original variables was gradient-colored related to its coordinate (coord) values on these two axes. b immunological phenotyping, PCA-derived latent variables, in connection with LVEF categorized by tertiles. Displayed as graph of the individuals between-class analysis (bca) of PCA was calculated from the same 19 variables in the sampled patients (n = 34). So dually, the diagram depicted the first principal plan represented by 28% and 17% of whole inertia for axis one (horizontal) and second (vertical) axis, respectively. The between-class inertia percentage was 11% (p = 0.022). Categorization in tertile 1 [28–55] (n = 12), tertile 2 [55–67] (n = 12), and tertile 3 [67–80] (n = 10) of LVEF [%] was used as the instrumental variable and colored-labeled for each patient individual (point). The observed probability resulted from a Monte Carlo permutation test done on this between-class inertia