Skip to main content
. Author manuscript; available in PMC: 2023 Feb 1.
Published in final edited form as: Toxicol In Vitro. 2022 Nov 6;86:105515. doi: 10.1016/j.tiv.2022.105515

Table 5 A.

OptiSafe GHS NC vs. Rest. B. Updated OptiSafe GHS NC vs. Rest. C. OptiSafe GHS Cat. 1 vs. Rest. D. Updated OptiSafe GHS Cat. 1 vs. Rest.

GHS NC vs. Rest
A. OptiSafe
B. Updated OptiSafe
Statistic OptiSafe Statistic Updated OptiSafe
Sensitivity 100.0% (88/88) Sensitivity 100.0% (89/89)
Specificity 62.5% (30/48) Specificity 79.6% (39/49)
FN Rate 0.0% (0/88) FN Rate 0.0% (0/89)
FP Rate 37.5% (18/48) FP Rate 20.4% (10/49)
Accuracy 86.8% (118/136) Accuracy 92.8% (128/138)
Bal. Accuracy 81.3% Bal. Accuracy 89.8%
GHS Cat.1 vs Rest
C. OptiSafe
D. Updated OptiSafe
Statistic OptiSafe Statistic Updated OptiSafe

Sensitivity 82.2% (37/45) Sensitivity 91.7% (44/48)
Specificity 67.9% (55/81) Specificity 71.8% (56/78)
FN Rate 17.8% (8/45) FN Rate 8.3% (4/48)
FP Rate 32.1% (26/81) FP Rate 28.2% (22/78)
Accuracy 73.0% (92/126) Accuracy 79.4% (100/126)
Bal. Accuracy 75.1% Bal. Accuracy 81.8%

Cat. = Category; FN Rate = False-negative rate; FP Rate = False-positive rate; Bal. Acc. = Balanced Accuracy; NC = Not classified. Table 4 explains the differences in n. GHS NC versus the rest used 2/1 predictions and GHS NC versus the rest does not use 2/1 predictions. See text for details.