A Measure Evaluating Relevance of a Validation Study of Alternatives to Animal Testing

抄録

Sensitivity, specificity and accuracy are well known measures for evaluating the relevance of an inter-laboratory validation study for alternative tests. It is not generally discussed that the measures are dependent on two determining factors: a set of chemicals and the number of laboratories. Furthermore, some alternative tests such as these for the phototoxicity test have an "Equivocal" category for judging the toxicity of chemicals. These facts have made it difficult to interpret the value of the measures.
Therefore, in this paper we propose new measures to evaluate the alternatives, which depend on a set of chemicals rather than on both factors, and can treat data which have "Equivocal" category. We also propose their confidence intervals, which are measures of their precision.