Skip to main content

Table 7 Tool performance and comparison summary statistics to 96 h-LC50 of fish based on entire dataset

From: Comparison of seven in silico tools for evaluating of daphnia and fish acute toxicity: case study on Chinese Priority Controlled Chemicals and new chemicals

Chemicals Methods Measures of predictive accuracy ECOSAR T.E.S.T Danish Q.D. VEGA Read Across Trend Analysis KATE
37 PCCs Number of missing predictions 0 2 3 1 2 4 9
Qualitative Number of correct 20 17 17 17 14 10 10
Number of incorrect 17 18 17 19 21 23 18
Total accuracy (%)a 54 49 50 47 40 30 36
Predictive power (%)b 54 46 46 46 38 27 27
R2 (toxicity class) 0.50 0.39 0.38 0.43 0.17 0.10 0.25
Quantitative Accuracy within a factor of 10 (%) 68 80 65 81 57 36 71
Accuracy within a factor of 100 (%) 89 89 79 94 83 48 86
Accuracy within a factor of 1000 (%) 92 97 85 97 94 76 89
R2(log10 LC50) 0.31 0.35 0.27 0.34 0.32 0.03 0.21
86 NCs Qualitative Number of correct 34 25 17 18 / / 17
Number of incorrect 47 36 24 65 / / 40
Number of missing predictions 5 25 45 3 / / 29
Total accuracy (%)a 42 41 41 22 / / 30
Predictive power (%)b 40 29 20 21 / / 20
R2 (toxicity class) 0.08 0.10 0.13 0.001 / / 0.03
  1. a Total accuracy is the fraction of chemicals assessed by each tool for which the predicted LC50 falls within the same regulatory category as the measured LC50. b Similar to total accuracy, predictive power measures the total number of correct category assignments. However, lack of prediction was treated as an incorrect assignment