Skip to main content
. 2023 Aug 24;6:870. doi: 10.1038/s42003-023-05243-w

Fig. 4. Evaluation of our method on the new drugs.

Fig. 4

a Mean AUC and AUPR at predicting the drug–ADR interactions that cause serious clinical outcomes and the corresponding classes of serious clinical outcomes for varying percentages of randomly chosen interactions. b The distribution of AUCs and AUPRs of drugs in de novo test. Since some drugs only have one or more of the seven seriousness classes in the benchmark dataset, it is not possible to obtain all the seriousness class labels. The boxes inside the violin plots indicate the 50-th percentile (middle line), 25-th and 75-th percentile (box), the smallest value within 1.5 times interquartile range below the 25-th percentile, and the largest value within 1.5 times interquartile range above the 75-th percentile (whiskers), and outliers (dots). We set the AUC and AUPR values of the corresponding seriousness classes of these samples to 0 and do not count them.