Skip to main content

Table 1 Model evaluation scores for all datasets. First rank scores and AttentionDDI model scores are reported in bold

From: AttentionDDI: Siamese attention-based deep learning method for drug–drug interaction predictions

Model score DS1 DS2 DS3 (CYP) DS3 (NCYP)
  AUC AUPR AUC AUPR AUC AUPR AUC AUPR
AttentionDDI\(\ddag\) 0.954 0.924 0.986 0.904 0.989 0.775 0.986 0.890
AttentionDDI (without siamese)\(\ddag\) 0.944 0.907 0.965 0.791 0.945 0.277 0.907 0.443
AttentionDDI (without Attention & siamese)\(\ddag\) 0.944 0.909 0.926 0.596 0.962 0.491 0.953 0.639
NDD* 0.954 0.922 0.994 0.890 0.994 0.830 0.992 0.947
 Classifier ensemble* 0.956 0.928 0.936 0.487 0.990 0.541 0.986 0.756
 Weighted average ensemble* 0.948 0.919 0.646 0.440 0.695 0.484 0.974 0.599
 RF* 0.830 0.693 0.982 0.812 0.737 0.092 0.889 0.167
 LR* 0.941 0.905 0.911 0.251 0.977 0.487 0.916 0.472
 Adaptive boosting* 0.722 0.587 0.904 0.185 0.830 0.143 0.709 0.150
 LDA* 0.935 0.898 0.894 0.215 0.953 0.327 0.889 0.414
 QDA* 0.857 0.802 0.926 0.466 0.709 0.317 0.536 0.260
 KNN* 0.730 0.134 0.927 0.785 0.590 0.064 0.603 0.235
ISCMF\(\dag\) 0.899 0.864 0.898 0.767 0.898 0.792
 Classifier ensemble\(\dag\) 0.957 0.807 0.990 0.541 0.986 0.756
 Weighted average ensemble\(\dag\) 0.951 0.795 0.695 0.484 0.974 0.599
 Matrix perturbation\(\dag\) 0.948 0.782
 Neighbor recommender\(\dag\) 0.953 0.126 0.904 0.295
 Label propagation\(\dag\) 0.952 0.126
 Random walk\(\dag\) 0.895 0.181
  1. \(\ddag\)Our model, *scores from [8], \(\dag\)scores from [9]