LAUSR

Advances in non‐clinical in vitro models, higher throughput approaches and the promise of human‐derived preparations require methods to reliably assess the fidelity of translation of such assays, compared with in vivo models and clinical studies. This review discusses general principles and parameters useful to evaluate the value of non‐clinical assays typically used to guide compound progression. I first consider the biological characteristics (including sensitivity and ability to replicate relevant responses) of models that form the foundation of an assay based on the questions posed. I then discuss the quantitative assessment of diagnostic performance and assay utility, including sensitivity and specificity, receiver operating characteristic curves, positive and negative predictive values, likelihood ratios, along with advantages of combining two independent assays. Understanding the strengths and limitations of the biological model employed along with assay performance and context of use is essential to selecting the best assays supporting the best drug candidates.