Skip to main content

Advertisement

Table 8 External performance of continuous models for predicting classification endpoints (vT, nT, EPA, GHS)

From: SAR and QSAR modeling of a large collection of LD50 rat acute oral toxicity data

  SEN SPE MCC BA #AD %AD CS
nT 0.794 0.796 0.587 0.795 2665 0.924 1
0.840 0.841 0.677 0.840 2182 0.757 2
0.878 0.858 0.733 0.868 1704 0.591 3
0.913 0.883 0.794 0.898 1222 0.424 4
vT 0.743 0.938 0.577 0.840 2742 0.949 1
0.796 0.976 0.737 0.886 2316 0.802 2
0.890 0.978 0.820 0.934 1556 0.539 3
EPA 0.602 0.856 0.439 0.729 2653 0.928 1
0.701 0.885 0.550 0.793 1731 0.605 2
0.739 0.898 0.600 0.819 1200 0.420 3
GHS 0.567 0.894 0.461 0.731 1561 0.542 1
0.644 0.911 0.541 0.777 908 0.315 2
0.676 0.916 0.573 0.796 617 0.214 3
  1. For each model, the sensitivity (SEN), the specificity (SPE), the balanced accuracy (BA), the Matthew’s correlation coefficient (MCC) the number (#AD) and the percentage (%AD) of predictions in AD are reported, with respect to the CS threshold for defining predictions in AD. For multi-category endpoints (EPA and GHS), SEN and SPE are the average of sensitivities/specificities computed separately for each class, while BA is the arithmetic mean of the average SEN and SPE