value proposition, to drug discovery and preclinical/clinical drug development. The exposition varies from general description of clinical trials fundamentals (such as placebo effect, blinding, and randomization), to somewhat detailed descriptions of certain topics (e.g., the K-arm pick-winner design). A clinical trial is seen as one component of a clinical development program, and therefore the goal is to design what authors call “globally optimal” studies that maximize the expected gain (typically monetary gain). An interesting example on chronic obstructive pulmonary disease (COPD) illustrates the main ideas. The remaining chapters address topics common in clinical trials. They are independent of each other, and can be used as brief overviews of those particular topics. Briefly, Chapter 5 describes trial designs for precision medicine. Various issues related to prognostic/predictive biomarkers are discussed, along with various study designs informed by these biomarkers (e.g., biomarker-stratified, and biomarker-adaptive designs). Brief descriptions of umbrella and basket designs are also included. Time-to-event outcomes are discussed in Chapter 6. Standard tools (e.g., the Nelson–Aalen and Kaplan–Meier estimators, and the Cox proportional hazards model) for such outcomes are only briefly described. Instead, the authors introduce more advanced topics, such as landmarking, delayed drug effects, treatment switching, and competing risks. Chapter 7 describes statistical issues in multiple-hypothesis testing. The material covered ranges from standard union-intersection and intersection-union tests, to more modern topics such as the error-average method, the Li–Huque’s method, and analysis of composite outcomes using prioritized endpoints. Statistical techniques for missing data (including multiple imputations and inverse-probability-weighting methods) are included in Chapter 8, while a very short introduction to the role played by data monitoring committees in clinical trials is provided in Chapter 10. Chapter 9 (Special Issues and Resolutions) provides very brief descriptions for a number of different topics (e.g., drop-the-loser designs based on efficacy and safety, a comparison of the Fisher’s vs. Bernard’s exact tests), without a unifying theme connecting them. Chapter 11 describes a number of issues labeled as Controversies in Statistical Science. Although interesting, it is not immediately clear what is really controversial about some of the topics included (such as regression with time-dependent variables). The book is generally well written. However, there is a relatively large number of typos and grammatical errors, as well as some minor slips (e.g., the test statistic T2 in Equation (4.2) on page 89 is a linear combination of z-statistics, so its distribution cannot be “uniform on [0,1] under null hypothesis H0,” as stated 7 lines below Equation (4.2)). But this is mostly nitpicking. More details added to the figures and tables captions would be helpful to a casual reader. The book has a number of detailed examples, and SAS and R code to implement some of the methods described in the text, but no exercises or datasets. While it was not meant to be used as a textbook, some of the chapters (such as Chapters 1–4) certainly deserve to be on the recommended readings list for a graduate level class on clinical trials. In addition, most chapters can be used as (approximately 30-page long) overview/reference for the topics they cover. While some of the material in the book may be accessible to a wider audience, most of the specific topics described would likely require a graduate level background in Biostatistics or a related field. In summary, this book covers a wide range of interesting topics in clinical trials, and provides an appealing and useful reference to researchers.

中文翻译：

---

基于模型的全球公共卫生地质统计学：方法和应用

价值主张，用于药物发现和临床前/临床药物开发。阐述从临床试验基本原理的一般描述（如安慰剂效应、盲法和随机化）到某些主题的详细描述（例如，K 臂选择优胜者设计）不等。临床试验被视为临床开发计划的一个组成部分，因此目标是设计作者所说的“全局最优”研究，以最大化预期收益（通常是货币收益）。一个关于慢性阻塞性肺病 (COPD) 的有趣例子说明了主要思想。其余章节涉及临床试验中常见的主题。它们彼此独立，可以用作这些特定主题的简要概述。简而言之，第 5 章描述了精准医学的试验设计。讨论了与预后/预测性生物标志物相关的各种问题，以及由这些生物标志物提供的各种研究设计（例如，生物标志物分层和生物标志物适应性设计）。还包括雨伞和篮子设计的简要说明。事件发生时间结果在第 6 章中讨论。此类结果的标准工具（例如，Nelson-Aalen 和 Kaplan-Meier 估计量，以及 Cox 比例风险模型）仅简要描述。相反，作者介绍了更高级的主题，例如标志性、延迟药物效应、治疗转换和竞争风险。第 7 章描述了多假设检验中的统计问题。涵盖的材料范围从标准的联合-相交和相交-联合测试，到更现代的主题，如误差平均法、李胡雀法、和使用优先终点分析复合结果。缺失数据的统计技术（包括多重插补和逆概率​​加权方法）包含在第 8 章中，而第 10 章提供了对数据监测委员会在临床试验中所发挥作用的非常简短的介绍。 第 9 章（特殊问题）和决议）为许多不同的主题提供了非常简短的描述（例如，基于有效性和安全性的失败者设计，Fisher's 与 Bernard's 精确测试的比较），没有将它们连接起来的统一主题。第 11 章描述了许多标记为统计科学中的争议的问题。尽管很有趣，但目前还不清楚关于所包含的某些主题（例如使用时间相关变量的回归）真正有争议的是什么。这本书总体上写得很好。但是，错别字和语法错误比较多，还有一些小错误（例如第 89 页公式（4.2）中的检验统计量 T2 是 z 统计量的线性组合，因此其分布不可能是“在零假设 H0 下在 [0,1] 上均匀”，如等式 (4.2) 下方的 7 行所述）。但这主要是吹毛求疵。添加到图形和表格标题中的更多细节对普通读者会有所帮助。书中有许多详细的例子，以及SAS和R代码来实现文中描述的一些方法，但没有习题或数据集。虽然它不打算用作教科书，但其中一些章节（例如第 1-4 章）当然值得列入临床试验研究生级别课程的推荐阅读清单。此外，大多数章节都可以用作（大约 30 页长）它们所涵盖主题的概述/参考。虽然本书中的某些材料可供更广泛的读者访问，但所描述的大多数特定主题可能需要生物统计学或相关领域的研究生水平背景。总之，本书涵盖了临床试验中广泛的有趣主题，为研究人员提供了有吸引力和有用的参考。