Skip to main content
. Author manuscript; available in PMC: 2019 Feb 1.
Published in final edited form as: Comput Toxicol. 2017 Apr 28;5:25–30. doi: 10.1016/j.comtox.2017.04.001

Table 2. Additional performance metrics for the top three teams in sub-challenge 2 of the sbv IMPROVER Systems Toxicology Challenge.

The metrics were computed using a cut-off of 0.5 on the confidence values that a sample was taken from a current smoker (left) or former smoker (right). For the former smoker vs never-smoker comparison (right) performance metrics were obtained from all samples and hence correspond to values in parentheses in Table 1.

Smoker vs Non-Current Smoker Former Smoker vs Never-smoker
Team Sensitivity (%) Specificity (%) Balanced accuracy (%) Sensitivity (%) Specificity (%) Balanced accuracy (%)
219 85 93 89 74 39 56
250 90 79 84 59 46 53
264 74 85 80 85 32 59