Skip to main content
. 2019 Dec 18;23(1):100780. doi: 10.1016/j.isci.2019.100780

Figure 4.

Figure 4

Accuracy of AML Classification in Different Leukemia Types and AML Subclasses

(A) Schema for determining accuracy for leukemia types and AML subclasses in dataset 1.

(B–D) Normalized dataset 1 was randomly split into training and test sets 100 times (same permutations as in Figure 2B), and prediction accuracy is reported for each individual sample. The bars in the figure correspond to individual samples broken down by leukemia type (B) and AML subtype (C). (D) Schema for determining accuracy for leukemia types and AML subclasses in dataset 2.

(E–H) Normalized dataset 2 was randomly split into training and test sets 100 times (D) and prediction accuracy is reported for each individual sample, listed by leukemia type (E) and AML subtype (F). Workflow for M3 subtype prediction using dataset 2 (G) Boxplots of prediction accuracy, sensitivity, and specificity over 100 train/test splits (H). Error bars depict the standard deviation.