ADTree predictive model of aGVHD. The ADTree consists of various levels of prediction (ellipses) and decision nodes (rectangles). Each prediction node is associated with a weight, representing its contribution to the cumulative prediction score (the larger scores indicate higher incidence of aGVHD), whereas each decision node contains a splitting attribute. Variables are not mutually exclusive. Patients traverse the tree according to their pretransplant characteristics (ie, variable values), and the cumulative scores for grade II-IV aGVHD (A) and grade III-IV aGVHD (B) are calculated. Numbers that appear before the variable names indicate the rankings in ADTree. AA, aplastic anemia; AML/MDS, acute myeloid leukemia or myelodysplastic syndrome; ATL, adult T-cell leukemia/lymphoma; BMI, body mass index; Dx, diagnosis; M to F, male donor to female recipient; Rel-BM, bone marrow from relatives; NHL, non-Hodgkin lymphoma; Pt, patient; Rel-PB, peripheral blood stem cell from relatives; RIC, reduced-intensity conditioning regimen; SCT, stem cell transplantation; Tac, tacrolimus; TBI, total body irradiation; UR-BM, bone marrow from unrelated donor; yo, years old.