Competing Failure Directional Swimming In The Design Of The Optimization Of The Dose-response Mode

This paper emphasizes on the study of allocation designs for the prob lem of determining the optimal therapeutic dose for subjects in early-phase clinical trials. It draws lessons from the advantage and disadvantage of the Directed Walk Algorithm Design and its application in allocation, aiming at the question of how to determine the next dose when there are three types of responses whose solution is still up in the air. Based on the basic basic idea of Directed Walk Algorithm Design, we propose a new design which combines with the optimizing up-and-down design . In classical dose-response problems, a subject's response is modeled as a Bernoulli random variable , that is , toxic, nontoxic.Now we separate it into three levels:heavy toxic, light toxic, nontoxic.When there are heavy toxic or nontoxic, we use up-and-down design. When light toxic, we use directed walk design.This design improves the decision efficiency and is more safe.This paper studies a competing failure model in which the goal is to maximize the probability that a subject being treated exhibits both a nontoxic response and is cured.We know, a subject can fail due to a toxic reaction or due to lack of efficacy.Successful subjects will have both a positive response and no toxic side effects.Thus , we seek to maximize the product of the nontoxicity and efficacy dose-response.In this paper, simulation is used as a design aid.And we compare this new design with previous design. At last, this paper looks forward to the future development of this field.