Fig. 1.

Our framework with the three principal steps: (1) extraction of the raw features (from MRIs, i.e. $v_1,\dots ,v_s$, and from clinical data, i.e. $c_1,\dots ,c_t$); (2) evaluation of the important information by using EasyMKLFS for feature weighting and selection; (3) generation of the final sparse model.