Classification model accurately predicts B-ALL subtypes
(A) Flow chart outlines process for PCA-LDA classification of B-ALL subtypes.
(B) Confusion matrix showing number (listed) and percentage (color coded) of B-ALL subtype truths and predictions for leave-one-out cross-validation.
(C) Three-dimensional plots showing clustering of B-ALL subtypes utilizing select dimensions from the LDA model.
(D) LDA clustering of validation cohort B-ALL patient samples with training samples after processing with the classification algorithm. Validation samples for each subtype group are shown in black. Data are provided for DUX4-rearranged (left), ETV6::RUNX1 (center), and hyperdiploid (right) B-ALL.