Innovative Clinical Trial Design and Analysis

研究设计是临床试验中至关重要的第一步。精心设计的研究对于成功的癌症至关重要 研究和癌症药物开发。创新的临床试验设计可能需要更少的 患者，节省资源，加速癌症药物开发。这一广泛而长期的目标 研究项目是开发新的统计方法，以解决新的和具有挑战性的问题 癌症临床试验的设计和分析。本项目有三个具体目标。第一个目标是 纵向数据和关节模型的设计和样本量计算的统计方法 纵向和生存数据。将开发统计方法，用于样本量和功效估计， 治疗对生存期的总体和直接影响，纵向过程对生存期的影响，以及 用于涉及多变量纵向和多变量生存过程的设置。第二个目的研究 有生存率的癌症预防随机分组试验设计和分析的统计方法学 和复发事件结果。经验过程理论将被用来研究的渐近行为的 将使用检验统计量和渐近近似以及排列检验来开发样本 大小公式和功率估计。第三个目标是解决肿瘤学中的重要统计问题 药物开发途径。有三个次级目标。第一个分目标是在有针对性的设计领域。 将开发替代设计的方法，包括"浓缩"设计， 将比较这些设计与完全目标设计的特性和成本。有效和高效 这些试验的统计方法将通过应用半参数经验似然来开发 approach.第二个分目标是在第11阶段设计领域。II期和11/111期的新方法 将开发临床试验，并研究其操作特性、成本和对后续研究的预测能力。 将评估III期试验。11期研究中联合和非联合治疗的信息 研究和后续的第三阶段研究将被收集，以使用机器学习建立预测模型 以及其他非参数分类方法。第三个子目标是部分随机化领域 的设计.新的半参数经验似然方法将被开发用于设计和分析 这样的试验，以调整选择偏差，提高效率。我们的研究将产生重要的新的和 癌症研究的高效设计和分析工具。

Study design is a crucial first step in clinical trials. Well-designed studies are essential for successful cancer research and cancer drug development. Innovative clinical trial designs can potentially require fewer patients, save resources, and accelerate cancer drug development. The broad, long-term objective of this research project is to develop new statistical methodology to address new and challenging issues in the design and analysis of cancer clinical trials. There are 3 specific aims in this praject. The first aim addresses statistical methods for the design and sample size calculation for longitudinal data and joint models for longitudinal and survival data. Statistical methods will be developed for sample size and power estimation for the overall and direct treatment effect on survival, for the effect of the longitudinal process on survival, and for settings involving multivariate longitudinal and multivariate survival processes. The second aim studies statistical methodology for the design and analysis of group randomized cancer prevention trials with survival and recurrent event outcomes. Empirical process theory will be used to study the asymptotic behavior of the test statistics and both asymptotic approximation as well as permutation test will be used to develop sample size formulas and power estimation. The third aim addresses important statistical issues in the oncology drug development pathway. There are three sub-aims. The first sub-aim is in the area of targeted designs. Methods for alternative designs, including "enrichment" designs, will be developed, and the operating characteristics and costs of these designs to fully targeted designs will be compared. Valid and efficient statistical methods for these trials will be developed by applying a semiparametric empirical likelihood approach. The second sub-aim is in the area of phase 11 designs. New methods for phase II and phase 11/111 clinical trials will be developed and their operating characteristics, costs, and predictive ability for subsequent phase HI trials will be assessed. Information on both combination and non-combination therapies in phase 11 studies and subsequent phase III studies will be gathered to build prediction models using machine learning and other nonparametric classification methods. The third sub-aim is in the area of partially randomized designs. New semiparametric empirical likelihood methods will be developed for the design and analysis of such trials to adjust for selection bias and to improve efficiency. Our research will produce important new and efficient design and analysis tools for cancer research.