Algorithms to generate designs of potency experiments that use far fewer animals

 描述(由申请人提供)：在动物身上进行的相对效力研究，以确定对抗辐射致死的药物和预防炭疽等致命疾病的疫苗的有效性，通常会使用比必要更多的动物。这似乎是因为缺乏适当的工具和指导来有效地设计这种研究，尽管人们对减少使用动物有广泛的兴趣，主要是为了限制疼痛和 动物可能经历的痛苦(De Moura等人，2009；McFarland等人，2011)。鉴于人体研究用于预防致命性和致命疾病的制剂是不道德和不可行的，根据FDA的动物规则(FDA 2002,2009)，此类研究必须在动物身上进行，通常是在狗和非人类灵长类等大型动物身上进行。在这个项目中，我们将开发易于使用的设计工具，使研究人员能够在不牺牲统计能力的情况下，大幅减少他们在效力研究中使用的动物数量。该项目将建立在我们之前关于确定辐射致死性研究样本量的工作(Kodell等人，2010年；Landes等人，2013年)的基础上，在该工作中，我们发现，参照物和试验治疗交错剂量的实验设计比传统的相同剂量设计效率高得多。在这个项目中，我们将开发新的算法，研究人员可以使用这些算法来设计强大的交错剂量实验，将动物数量保持在最低限度，并从数量和剂量放置的角度来评估权衡。图形功能将允许用户直观地评估与相同剂量设计相比，使用交错剂量设计可以实现的动物数量节省，并评估错误指定的输入参数对功率的影响。我们将在Excel电子表格、SAS/IML宏和R包(CRAN)中免费提供算法以及指导，以提高进行相对效力研究的研究人员使用这些高效设计以尽可能少地使用动物来实现他们的目标的可能性。

 DESCRIPTION (provided by applicant): Relative potency studies conducted in animals to establish the efficacy of drugs to counter radiation-induced lethality and of vaccines to protect against deadly diseases like anthrax often use more animals than necessary. This appears to be due to the absence of appropriate tools and guidance for the efficient design of such studies, even though there is broad interest in reducing the use of animals, primarily to limit the pain and suffering the animals may experience (de Moura et al., 2009; McFarland et al., 2011). Given that human studies of agents to protect against lethality and deadly diseases are unethical and infeasible, such studies must be conducted in animals, often in large animals like dogs and nonhuman primates, in accordance with the FDA Animal Rule (FDA 2002, 2009). In this project we will develop easy-to-use design tools to enable researchers to reduce substantially the number of animals they use in potency studies, without sacrificing statistical power. The project will build on our previous work on sample-size determination for radiation lethality studies (Kodell et al., 2010; Landes et al., 2013), in which we discovered that experimental designs with staggered doses of reference and test treatments are far more efficient than traditional, same-dose designs. In this project we will develop novel algorithms that researchers can use to design powerful, staggered-dose experiments that keep animal numbers to a minimum, and to evaluate trade-offs in terms of numbers and placement of doses. Graphical features will allow users to evaluate visually the savings in animal numbers that can be achieved using staggered-dose designs compared to same-dose designs, and to evaluate the effect on power of misspecified input parameters. We will provide the algorithms at no cost in an Excel spreadsheet, SAS/IML macro, and R package (CRAN), along with guidance, to enhance the likelihood that researchers conducting relative potency studies will employ these efficient designs to use the fewest animals possible to meet their objectives.