Welcome to the Bayesian Statistical Science Section

The Bayesian Statistical Science Section of the ASA provides a forum for statisticians and people who have interest in the Bayesian paradigm. The broad objectives of the Section are: to encourage research on theory and methods of statistical inference and decisionmaking associated with Bayes' theorem and to encourage the application and proper use of Bayesian procedures in the behavioral, biological, managerial, engineering, environmental, legal, medical, pharmaceutical, physical, and social sciences.

Current Affairs 

Upcoming Web-Based Lectures

Title: Bayesian Methods in Pharmaceutical Development

Presenter: Mani Lakshminarayanan, Pfizer, Inc and Karen Price, Eli Lilly and Company
Date and Time: Wednesday, October 29, 12:00 p.m. - 2:00 p.m. Eastern Time
Sponsor: Section on Bayesian Statistical Science and the International Society for Bayesian Analysis

Registration Deadline: Monday, October 27, at 12:00 p.m. Eastern time

In this data-centric computer age, especially when the scientific community is baffled with how to deal with "Big Data" that has a potential to transform business, health care, scientific discovery and public policy in a significant manner, Bayesian methods are making inroads into the scientific community with rapid growth and acceptance. This upsurge of interest is also seen in various scientific areas, technology, social sciences, management and commerce, where the members have started recognizing the benefits and flexibility of Bayesian methods in supporting their scientific quest. In the pharmaceutical research, both NICE (National Institute of Clinical Excellence) and the Medical Devices Division of the Food and Drug Administration have already signaled their acceptance of Bayesian methods in their submissions and have taken big steps in providing relevant guidance on this topic. Various other organizations have indicated support of increasing the appropriate use of Bayesian methods, such as the Drug Information Association (DIA) which houses the Bayesian Scientific Working Group (DIA BSWG). The DIA BSWG has a vision to ensure that Bayesian methods are well-understood and broadly utilized for design, analysis and throughout the medical product development process and to improve industrial, regulatory and economic decision making.

In this webinar, our primary objective is to discuss some of the current topics in which Bayesian methods provide a natural platform to quantify and address directly key questions and issues that arise at all stages of pharmaceutical product development. Our discussion will fully highlight and elaborate on some of the areas that the DIA BSWG has been working on since 2011 including safety signal detection, network meta-analysis and non-inferiority trials. In addition, we will discuss topics such as use of probability of success and prior elicitation. Such topics are relevant in addressing a common issue that arises in projects involving clinical trial development, that is, how to integrate findings from early phase trials explicitly and other compounds' trials data in order to objectively provide specific quantitative guidance for the design of later trials. Our primary intention is to focus on discussing methodology and applications with examples that are prevalent currently in the pharmaceutical community.

*** More information and registration at the webinar page.

Title: Coordinating Collection, Analysis and Submission of Clinical Trials Data with Metadata Standards - Focus on Analysis Deliverables (Part 2 of 2)
Presenter: Steve Kirby, Mario Widel and Luke Reinbolt
Date and Time: Tuesday, November 4, 2014, 12 p.m. - 1:30 p.m. Eastern time
Sponsor: Section for Statistical Programmers and Analysts

Registration Deadline: Friday, October 31, at 12:00 p.m. Eastern time

Starting with a diagram that links key submission deliverables and associated metadata, the authors will share how companies can plan to have key content needed to support study planning and generation of CDISC data, TFLs and associated documentation available as metadata and will discuss how that metadata can support operational efficiency and a transparent, traceable path from study planning to regulatory submission.

For each deliverable, the authors will explore the use case for a metadata driven approach, share a specific, concrete process example and investigate how the content is related to other aspects of the clinical trials process.

Part 2 will focus on analysis deliverables and associated metadata. As what actually happens in a study (or related studies) impacts analysis, this analysis content is in some cases impacted by tabulations content presented in Part 1. Key content areas will be: ADaM metadata and ADaM mapping code; TFL metadata and TFL code; and associated submission documentation (define.xml, Reviewer's Guide).

*** More information and registration at the webinar page.

Title: Bayesian Combination Dose Finding: Concepts and Applications
Presenter: Simon Wandel (Cogitars)
Date and Time: Tuesday, November 18, 2014, 12:00 p.m. - 2:00 p.m. Eastern time
Sponsor: Biopharmaceutical Section

Registration Deadline: Friday, November 14, at 12:00 p.m. Eastern time

Particular interest in Bayesian model-based dose finding has been observed in recent years. Multiple publications of successful single-agent phase I Oncology studies undermine the related paradigm shift which happened at some Pharmaceutical companies already and is ongoing in others. However, nowadays single agent dose finding is not the only goal in early development; rather, combination dose finding starts to play an important role which poses new challenges to clinical teams and statisticians.

In this webinar, an introduction to the statistical concept of Bayesian model-based dose finding with a particular emphasis on the combination setting will be provided. Based on practical experience, a special section will be devoted to the selection of the starting dose combination where both, statistical modeling and clinical considerations play an important role. Real case studies will be used to illustrate concepts and methods along with potential pitfalls which should be avoided. Finally, based on a literature search and on personnel communication, an overview of the current trends and adaption in the United States and in Europe will be provided.

*** More information and registration at the webinar page.

​Title: From Sample Size Calculations to Clinical Trial Optimization
Presenters: Alex Dmitrienko, Quintiles, and  Gautier Paux, Servier
Date and Time: Thursday, December 4, 2014, 12:00 p.m. – 2:00 p.m. Eastern time
Sponsor: Biopharmaceutical Section

Registration Deadline: Tuesday, December 2, at 12:00 p.m. Eastern time

It is well known that sample size determination is one of the key aspects of designing a new clinical trial.  A standard sample size calculation approach generally pursues a simple goal of computing a quick estimate of the required number of patients or events.  Closed-form solutions are normally preferred, which often forces the trial’s sponsor to make simplifying assumptions. The clinical scenario evaluation (CSE) framework introduced in Benda et al (2010) provides an extended version of this basic “one-dimensional” approach.  Clinical scenario evaluation focuses on a broad “multi-dimensional” approach to a quantitative assessment of the operating characteristics of several candidate analysis methods under multiple candidate trial designs to arrive at a solution which is consistent with the trial’s clinical objectives and maximizes a relevant success criterion or utility function.
In this webinar we will introduce the key principles of clinical scenario evaluation in the context of Phase II and Phase III clinical trials and touch upon multiple related approaches.  We will discuss the general concept of clinical trial optimization aimed at identifying the configurations of applicable design scenarios and analysis strategies that lead to optimal performance.  We will also emphasize the importance of sensitivity assessments to ensure that an optimal clinical trial design is robust to reasonable deviations from the assumed parameter values.
The CSE approach will be illustrated using case studies based on a clinical trial with multiple objectives, clinical trial with a biomarker-driven design and an adaptive Phase II clinical trial.  Finally, we will introduce an R package (Mediana package) which was developed to provide a general software implementation of the CSE approach. The current version of the Mediana package supports a broad set of clinical trial designs and analysis methods, and we will discuss new features that will be added in the future.

*** More information and registration at the webinar page.

From the SBSS Mixer at JSM 2013