Fig. 4From: Multiple machine-learning tools identifying prognostic biomarkers for acute Myeloid LeukemiaScreening important genes related to the prognosis of AML patients based on recursive feature elimination (RFE) algorithm. Feature selection is performed using multiple functions in the R package caret (lrFuncs, IdaProfile, caretFuncs, and nbFuncs). (A) lrFuncs model identified 15 genes as the optimal characteristic genes (maximum accuracy = 0.9595). (B) 26 genes were identified as the optimal feature genes by the IdaProfile model (maximum accuracy = 0.9595); (C) caretFuncs model identified 12 genes as the optimal characteristic genes (maximum accuracy = 0.9853); (D) nbFuncs model identified 24 genes as the optimal feature genes (maximum accuracy = 0.8856)Back to article page