Abstract:
In this article we consider identification of maximum safe dose (MSD) in a dose response study for distribution-free endpoints. Where the maximum safe dose is the highest dose level that does not exceed the median toxicity of the zero dose by a predetermined margin. A nonparametric confidence set-based approach was proposed, that is we incorporate Mann-Whitney method with the partitioning principle in a step-down fashion for safety evaluation. A comprehensive study of the familywise error rate for our new procedure was compared with the dose response (DR) procedure via a Monte Carlo simulations. An example from preclinical trial in genetic toxicology was used for illustrative purpose
