Figure 3
From: HIDEEP: a systems approach to predict hormone impacts on drug efficacy based on effect paths
Performance comparison with two previous methods. The performance of HIDEEP (pink), RWDDI (light blue) and NBDTI (green) for 13 diseases was evaluated by the area under the receiver operating characteristic curve (AUROC). For this performance comparison, we adopt an evaluation strategy, 30 times of 10-fold cross validation used to evaluate NBDTI in the original study. HIDEEP, RWDDI and NBDTI are applied to five datasets consisting of a gold standard set and five unlabeled sets with ‘10 times’ for 13 diseases with 10 or more positive samples. Thus, each box includes 150 AUROC values for five datasets.
