Innovative Methods for Research and Trial Design using Value of Information

临床试验测试了医疗治疗在人体上的效果、风险和好处，在推荐新的治疗方法在加拿大医疗体系中使用之前，需要进行临床试验。由于临床试验既昂贵又耗时，必须进行适当的设计，以确保试验期间收集的信息将帮助医生和政策制定者为患者找到最佳治疗方法。信息价值(VOI)是一种提出的设计研究的方法，提供物有所值并可以选择最佳治疗。这确保了在临床试验期间收集的信息支持医生和政策制定者做出治疗决策，并最大限度地发挥临床试验的影响。目前，VOI比标准的研究设计方法更复杂、更耗时、更昂贵，这意味着它还没有在实践中使用。我之前的研究开发了新的统计方法，以减少使用VoI方法所需的时间。因此，该项目将在这项工作的基础上开发统计方法，继续降低在现实世界中使用VoI的复杂性。在短期内，这将集中在三个关键领域：i)将Voi应用于现实的临床试验，ii)使用Voi设计临床试验，根据试验期间收集的数据改变其行为，称为适应性试验，以及iii)在证据非常有限的情况下改进Voi的使用。由于目前的研究集中在减少计算VOI所需的时间，因此关于如何将这些方法应用于现实生活中的临床试验的研究有限。最近，Voi的指导方针强调了考虑更现实的试验的重要性，但在如何实现这一点方面存在知识空白，这一空白将得到填补。适应性试验通常比标准试验更有效、更道德、更有信息量，但它们需要更多的时间来设计，特别是在使用Voi方法时。该项目将开发新的统计方法来减少这一时间，并允许使用VOI来提高适应性试验的效率。最后，试验通常被设计成在以前只收集了有限数量的数据时收集更多的信息。理论上，VOI仍然可以在这些“数据匮乏”的环境中使用，但我们需要新的方法来理解有限的信息是如何影响决策的，这样VOI才能准确地确定研究的优先顺序。该项目的总体目标是增加VOI在临床试验设计中的使用。除了短期目标外，我还将通过i)培训高素质的专业人员使用Voi并开发新方法，ii)开发代码和指导以帮助其他研究人员使用Voi，以及iii)与同事合作在实践中实施Voi，从而将Voi转化为一种可行的试验设计方法。这将改善VOI在试验设计中的使用，确保研究与关键决策者相关，并显著增加研究资金投资的回报。

Clinical trials test the effects, risks and benefits of medical treatments in people and are required before new treatments can be recommended for use in the Canadian healthcare system. As clinical trials are expensive and time consuming, they must be designed properly to ensure that the information collected during the trial will help doctors and policymakers to find the best treatment for patients. Value of Information (VoI) is a proposed method to design research that offers value-for-money and can select the best treatment. This ensures that the information collected during clinical trials supports doctors and policymakers to make decisions about treatments and maximizes the impact of clinical trials. Currently, VoI is more complex, time consuming and costly than standard research design methods, which means that it has not been used in practice. My previous research developed new statistical methods to reduce the time required to use VoI methods. Thus, this project will build on that work to develop statistical methods that continue to reduce the complexity of using VoI in the real world. In the short-term, this will focus on three key areas; i) applying VoI to realistic clinical trials, ii) using VoI to design clinical trials that change their conduct based on data collected during the trial, known as adaptive trials, and iii) improving the use of VoI when there is very limited evidence. As current research has focused on reducing the time taken to compute VoI, there is limited research on how to apply these methods to real-life clinical trials. Recently guidelines for VoI have highlighted the importance of considering more realistic trials but there is a gap in knowledge about how this can be achieved, which will be filled. Adaptive trials are often more efficient, ethical and informative than standard trials but they take more time to design, particularly when using VoI methods. This project will develop new statistical methods to reduce this time and allow VoI to be used to improve the efficiency of adaptive trials. Finally, trials are usually designed to collect more information when only a limited amount of data has been collected previously. In theory, VoI can still be used in these "data-poor" settings but we need new methods to understand how limited information influences decision making so VoI can accurately prioritize research. The overall goal of this project is to increase the use of VoI in clinical trial design. Alongside the short-term objectives, I will translate VoI into a feasible method for trial design by i) training highly qualified professionals to use VoI and develop new methods, ii) developing code and guidance to help other researchers use VoI and iii) collaborating with colleagues to implement VoI in practice. This will improve the use of VoI in trial design, ensuring that research is relevant to key decision makers and significantly increasing returns from research funding investment.