We can and should take advantage of nonmonotonic properties to perform statistical analysis rigorously by new statistical and morphometric methods.
Abstract
We aimed to a) determine whether BPA showed effects on the developing rat mammary gland using new quantitative and established semiquantitative methods in two laboratories, b) develop a software tool for automatic evaluation of quantifiable aspects of the mammary ductal tree, and c) compare those methods. Conclusions: Both the semiquantitative and the quantitative methods revealed nonmonotonic effects of BPA. The quantitative unsupervised analysis used 91 measurements and produced the most striking nonmonotonic dose–response curves. At all time points, lower doses resulted in larger effects, consistent with the core study, which revealed a significant increase of mammary adenocarcinoma incidence in the stop-dose animals at the lowest BPA dose tested.
