Short course programme
Full-day courses
| Title: | An introduction to Bayesian non-parametrics for causal inference |
| Teacher: | Michael Daniels (Univ. of Florida, USA) |
| Co teachers: | Jason Roy (Rutgers Univ., USA) |
| Description: | |
| Bayesian nonparametric (BNP) methods can be used to flexibly model joint or conditional distributions, as well as functional relationships. These methods, along with causal assumptions, can be used with the g-formula for inference about causal effects. This general approach to causal inference has several possible advantages over popular semiparametric methods, including efficiency gains, the ease of causal inference on any functionals of the distribution of potential outcomes, and the use of prior information. Importantly, these BNP methods capture uncertainty, not just about the distributions and/or functions, but also about causal identification assumptions. In this workshop we review BNP methods and illustrate their use for causal inference in the setting of point treatments, mediation, and semi-competing risks. We present several data examples and discuss software implementation using R. The R code and/or packages used to run the data examples will be provided to the attendees at a specific github site associated with their recent (2023) research monograph on the topic. Attendees should have some familiarity with Bayesian inference and causal inference (though the course will briefly review both). | |
| Title: | Compatible and informative simultaneous confidence intervals for clinical trials |
| Teacher: | Werner Brannath (Univ. of Bremen, Germany) |
| Co teachers: | Martin Scharpenberg (Univ. of Bremen), Mouna Akacha (Novartis, Switzerland) |
| Description: | |
| Compatible and simultaneous confidence intervals in trials with multiple hypothesis are receiving considerable attention from both drug developers and regulatory authorities. The goal for this workshop is to provide an introduction to the current landscape and a deeper understanding of important concepts, issues and complications coming with interval estimation for simultaneous inference in clinical trials. The concepts and methodologies will be illustrated with graphical multiple tests. We will discuss the information loss associated with compatible confidence intervals (compared to single-step methods) and review an approach to overcome this issue. The course will consist of introductory presentations and hands-on sessions with R using the packages gMCP and informativeSCI and will address theoretical and practical aspects. No specific background knowledge on multiple testing is required. | |
| Title: | A practical introduction to simulating complex trial designs |
| Teacher: | Thomas Jaki (Univ. of Regensburg, Germany, & Univ. of Cambridge, UK) |
| Co teachers: | Dominique-Laurent Couturier, Pavel Mozgunov (both Univ. of Cambridge) |
| Description: | |
| This course provides a comprehensive and hands-on introduction to the use of Monte Carlo (MC) simulation studies for the evaluation of innovative clinical trial designs, particularly focusing on Multi-Arm Multi-Stage (MAMS) designs that allow for the dropping of experimental arms before the end of a trial – either for efficacy or futility – based on pre-specified efficacy and futility bounds, thereby allowing for a more optimal allocation of resources. Upon completion of this course, participants will be able to: • Understand the rationale and advantages of using adaptive clinical trial designs as well as the rationale and advantages of using MC simulation studies to evaluate them, • Design and plan simulation studies using a structured approach and choose relevant simulation scenarios, benchmarks, along with appropriate performance measures. • Code and execute simulation studies effectively, considering parallelization. • Analyze simulation results comprehensively, taking the MC standard errors into account. For the hands-on sessions, participants should bring their laptops with a recent version of R/RStudio installed, along with the packages MAMS and future.apply. Participants in the course should have a basic understanding of clinical trial design principles and statistics in general. No prior knowledge of adaptive designs is required. Additionally, a basic level of R programming is necessary; participants should be familiar with R data structures, such as vectors and data frames, and should be able to write basic functions, modify existing ones, and use for loops effectively. |
|
Half-day courses
| Title: | Bayesian borrowing in clinical trials: design choices, assessment of operating characteristics and reporting |
| Teacher: | Annette Kopp-Schneider (German Cancer Research Center) |
| Co teachers: | Silvia Calderazzo (German Cancer Research Center) |
| Description: | |
| This course is meant to improve the understanding of Bayesian clinical trial designs incorporating external information, as well as to give guidance on how to investigate and communicate their properties. Special focus will be placed on the underlying trade-offs between robustness to heterogeneity between current and external trial information and sample size/power gains. In particular, the course will provide: • an overview of the Bayesian approach and its use in clinical trial hypothesis testing and effect estimation; • a review of the main Bayesian robust external information borrowing approaches available in this context; • analytical results and relationships on how information borrowing impacts the operating characteristics of the trial, i.e., its potential advantages as well as risks; • guidance on simulation studies and graphical reports to improve interpretation and communication transparency of the trial design. The course will also comprise a practical session where participants will be asked to discuss the implementation of information borrowing and its impact through case-studies. Pre-requisites are basic knowledge of probability distributions and hypothesis testing concepts. • an overview of the Bayesian approach and its use in clinical trial hypothesis testing and effect estimation; • a review of the main Bayesian robust external information borrowing approaches available in this context; • analytical results and relationships on how information borrowing impacts the operating characteristics of the trial, i.e., its potential advantages as well as risks; • guidance on simulation studies and graphical reports to improve interpretation and communication transparency of the trial design. The course will also comprise a practical session where participants will be asked to discuss the implementation of information borrowing and its impact through case-studies. Pre-requisites are basic knowledge of probability distributions and hypothesis testing concepts. |
|
| Title: | Post-hoc inference for true discovery proportion: methods and applications in R |
| Teacher: | Anna Vesely (University of Bologna, Italy) |
| Co teachers: | Angela Andreella (University of Trento, Italy) |
| Description: | |
| The course introduces post-hoc inference for the True Discovery Proportion (TDP), with a focus on methods based on conditional resampling and their application to biological data. First, it provides an overview of resampling-based hypothesis testing and explores its advantages over parametric approaches. Subsequently, it describes the challenges of multiple testing, highlighting that this is a non-trivial extension of individual hypothesis testing and that different generalizations of type I error control are possible. As main topic, it illustrates some recent proposals for TDP inference, with a particular focus on All-Resolutions Inference (ARI; Rosenblatt et al., 2018), permutation-based ARI (Andreella et al., 2023), and permutation-based true discovery guarantee by sum tests (Vesely et al., 2023). The course consists of a combination of theory and practical application. It begins with a theoretical lecture to establish foundational concepts, followed by a hands-on session where participants can apply the methods using R. Participants should have a foundational understanding of hypothesis testing, including the concepts of test statistic, p-value, type I error, and statistical power. Additionally, they should have a working knowledge of R, including the ability to import data, manage data frames, install packages and use their functions. |
|