Keynote Speakers

Professor Guosheng Yin, The University of Hong Kong

TitleEmpower p-value a Bayesian interpretation to control type I error in Bayesian adaptive design



Contrary to broad criticisms on the use of p-value in evidence-based studies, we justify its utility and reclaim its importance from the Bayesian perspective. Under noninformative prior distributions, we establish an equivalence relationship between p-value and Bayesian posterior probability of the null hypothesis. For two-sided hypothesis tests with a point null, we recast the problem as a combination of two one-sided hypotheses along the opposite directions and establish the notion of a “two-sided posterior probability,” which is equivalent to the (two-sided) p-value. In contrast to the common belief, such an equivalence relationship empowers p-value an explicit Bayesian interpretation of how strong the data support the null. Bayesian adaptive designs are usually based on the posterior distribution of the parameter of interest and calibration of certain threshold for decision making. To effectively maintain the overall type I error rate in multiple tests, we propose the cutoff boundaries for the posterior probabilities by establishing a connection between the Bayesian and the frequentist group sequential methods.

Guosheng Yin is Chair in Statistics in Department of Mathematics at Imperial College London.  Previously, he was Patrick S C Poon Endowed Professor and Head of the Department of Statistics and Actuarial Science at University of Hong Kong.  He received Ph.D. in Biostatistics from University of North Carolina at Chapel Hill in 2003.  In 2003-2009, he worked as Assistant Professor and then Associate Professor in Department of Biostatistics at University of Texas M.D. Anderson Cancer Center.  He was elected as a Fellow of the American Statistical Association in 2013 and Fellow of the Institute of Mathematical Statistics in 2021.  He served as Associate Editor for Journal of American Statistical Association, Bayesian Analysis, Contemporary Clinical Trials etc.  He has published over 220 peer-reviewed papers in clinical trial methodology, Bayesian adaptive design, survival analysis, high-dimensional data, change-point analysis and machine learning as well as two books on clinical trial designs.



Xinyuan Song, Chinese University of Hong Kong

Xinyuan Song is a full professor and Chair of the Department of Statistics, Chinese University of Hong Kong. Her research interests are latent variable models, nonparametric and semiparametric modeling, Bayesian methods, statistical computing, and survival analysis. She serves/served as an associate editor for a number of international journals in Statistics and Psychometrics, including Biometrics, Electronic Journal of Statistics, The Canadian Journal of Statistics, Statistics and Its Interface, Computational Statistics and Data Analysis, Psychometrika,  and Structural Equation Modeling.



Tutorial Speaker

Professor Ming-Hui Chen, University of Connecticut

Professor Chen is currently Board of Trustees Distinguished Professor and Head of the Department of Statistics at the University of Connecticut (UConn). He was elected to Fellow of International Society for Bayesian Analysis in 2016, Fellow of the Institute of Mathematical Statistics in 2007, Fellow of American Statistical Association in 2005. He received the University of Connecticut AAUP Research Excellence Award, the UConn College of Liberal Arts and Sciences (CLAS) Excellence in Research Award, the UConn Alumni Association’s University Award for Faculty Excellence in Research and Creativity, and ICSA Distinguished Achievement Award. He has served as Executive Director of ICSA, President of ICSA, Chair of Eastern Asia Chapter of ISBA, President of New England Statistical Society, Representative from Districts 1-3, ASA Caucus of Academic Representatives, and the 2022 JSM Program Chair.



Professor Yingying Wei, Chinese University of Hong Kong

TileBayesian Inference for Time-to-event Data on a Social Network

AbstractWith the rapid development of social media and electronic commerce, many social networks now encompass a sequence of time-to-event data for a pair of social actors. For example, in an email network, how long it takes for the receiver of an email to reply to the sender is available; in a company that uses web-based electronic document management systems such as those provided by Dropbox and ParaDM, the time an employee processes a file shared by another employee can be tracked. However, despite the active research on modeling the sending behaviors of messages in social networks with point processes, statistical research on response times and response rates on a social network is still lacking. The patterns of response times and response rates are highly heterogeneous. Even for the same person, his or her response speed can be very different when communicating with different people. Unfortunately, despite the large number of social network users, we often observe very sparse data. To deal with the challenges of nonresponse events, severe heterogeneity, and sparse observations, we propose a Bayesian hierarchical model—the survival mixed membership blockmodel (SMMB). The SMMB can simultaneously learn the network structure from the data and use the learned structure to improve actor pair-specific inference.  Because time-to-event data encode more information than binary data, we are able to prove that the SMMB is identifiable up to label switching without the pure node assumption required for the traditional mixed membership stochastic blockmodel. In this tutorial, we will first review the Bayesian inference for time-to-event data, in particular the Bayesian semiparametric cure rate model, and then discuss the SMMB and its inference.


Dr. Wei is an Associate Professor in the Department of Statistics at the Chinese University of Hong Kong. She obtained her bachelor’s degree in mathematics from Tsinghua University in 2009 and her MSc Eng degree in Computer Science and PhD degree in Biostatistics from Johns Hopkins University in 2014. Her main research interest lies in developing computationally efficient statistical methods to analyze noisy, complex and heterogeneous data arising from various cutting-edge fields such as genomics and social networks. Her paper on meta-clustering of genomic data received the W. J. Youden Award in Interlaboratory Testing from the American Statistical Association in 2019.