Bernard G. Greenberg Distinguished Lecture Series with Professor Amy H. Herring

The Department of Biostatistics is pleased to return to in-person lectures on the UNC Campus this year. Please join them in the Blue Cross Blue Shield Memorial Auditorium, 0001 Michael Hooker Research Center. A zoom option is provided below for those unable to attend in person.

Professor Amy H. Herring

Sara & Charles Ayres Professor

Statistical Science, Global Health, and Biostatistics & Bioinformatics

Duke University

Lecture 1: May 18, 2023, 10:00 a.m. – 11:00 a.m.

Bayesian Matrix Completion for Hypothesis Testing

Zoom Link:

https://unc.zoom.us/j/94181351351?pwd=SlpRcFBrN1JMaW5pQ2QzcWl1TStPQT09

Lecture 2: May 18, 2023, 2:00 p.m. – 3:00 p.m.

Bayesian Functional Factor Analysis using Nearly Mutually Orthogonal Processes

Zoom Link:

https://unc.zoom.us/j/98145056774?pwd=QUhrbHZqakxKT2Rsdm5jQnhxRCtiUT09

Lecture 3: May 19, 2023 10:00 a.m. – 11:00 a.m.

Informative Priors for Clustering

Zoom Link:

https://unc.zoom.us/j/96905158473?pwd=UzhlMVZsaGt0QmRyUUFUV0lIU25ZZz09

Lecture 1. Bayesian Matrix Completion for Hypothesis Testing

High-throughput screening (HTS) is used to rapidly and efficiently screen thousands of chemicals for potential toxicity. Testing using HTS primarily aims to differentiate active vs inactive chemicals for different types of biological assay endpoints across different species and cell types. However, even using high-throughput technology, it is not feasible to test all possible combinations of chemicals and assay endpoints, resulting in a majority of missing combinations. Our goal is to derive posterior probabilities of activity for each chemical by assay endpoint combination, addressing the sparsity of HTS data, with a focus on identifying chemicals potentially harmful to humans.  We can view this as a matrix-structured multiple hypothesis testing problem. We propose a Bayesian hierarchical framework, which borrows information across different chemicals and assay endpoints in a low-dimensional latent space. This framework facilitates out-of-sample prediction of bioactivity potential for new (chemical, assay endpoint) pairs not yet tested. Furthermore, we make a novel attempt in toxicology to model heteroscedastic errors as well as a nonparametric mean function, leading to a broader definition of chemical activity, whose need has been suggested by toxicologists. Application to data from EPA’s ToxCast and Tox21 data sets identifies chemicals that are most likely active in pathways relevant for two disease outcomes: neurodevelopmental disorders and obesity.

Lecture 2.  Bayesian Functional Factor Analysis using Nearly Mutually Orthogonal Processes

Functional factor analysis is an important dimension reduction method for functional and longitudinal data. Factor loadings give insight into patterns of variability of the observations, while latent factors provide a low-dimensional representation of the data that is useful for inferential tasks. Constraining the functional factor loadings to be mutually orthogonal is desirable for modeling but is computationally challenging. In this work, we introduce nearly mutually orthogonal processes, which can be used to effectively enforce mutual orthogonality of the factor loadings, while maintaining computational simplicity and efficiency. The joint distribution is governed by a penalty parameter that determines the degree to which the processes are mutually orthogonal and is related to ease of posterior computation. We demonstrate that our approach can be used for flexible and interpretable inference in an application to studying the effects of breastfeeding status, illness, and demographic factors on weight dynamics in early childhood.

Lecture 3.  Informative Priors for Clustering

Based on challenges in a large national study of birth defects, we consider a canonical problem in epidemiology of “lumping” versus “splitting” of groups. In many cases, groups may be unknown in advance, adding the additional challenge of determining group or cluster membership.  While there is a very rich literature proposing Bayesian approaches for clustering starting with a prior probability distribution on partitions, most approaches assume exchangeability. Even though there have been some proposals to relax the exchangeability assumption, allowing covariate-dependence and partial exchangeability, limited consideration has been given on how to include concrete prior knowledge on the partition itself. For example, we are motivated by an epidemiological application, in which we wish to cluster birth defects into groups and we have prior knowledge of an initial clustering, provided by experts. As a general approach for including such prior knowledge, we propose a Centered Partition (CP) process. Some properties of the CP prior are described, a general algorithm for posterior computation is developed, and we illustrate the methodology through simulation examples and an application to the motivating epidemiology study of birth defects.

Biography: Amy H. Herring is the Sara & Charles Ayres Distinguished Professor of Statistical Science, Global Health, and Biostatistics and Bioinformatics at Duke University. Dr. Herring received her doctorate in biostatistics at Harvard University in 2000 and joined the biostatistics faculty at UNC-Chapel Hill that same year. She was an esteemed faculty member at UNC, rising to the level of Distinguished Professor and Associate Chair of Biostatistics, before moving to Duke University in 2017. Her research interests include development of statistical methodology for longitudinal or clustered data, Bayesian methods, latent class and latent variable models, missing data, complex environmental mixtures, and applications of statistics in population health and medicine. She has received numerous awards for her work, including the Mortimer Spiegelman Award from the American Public Health Association as the best applied public health statistician under age 40. Her research program is funded by NIH, and she holds leadership positions at the national and international level, including as Chair of the American Statistical Association’s Section on Bayesian Statistical Science, as President-Elect of the International Society for Bayesian Analysis, and as a member of the Board of the International Biometric Society. She sits on several national committees, including the National Academy of Science Committee on Applied and Theoretical Statistics and the Research Committee of the Health Effects Institute.

The Bernard G. Greenberg Distinguished Lecture Series honors the first chair of the UNC Biostatistics Department, Dr. Bernard G. Greenberg, who later served with distinction as dean of the School of Public Health from 1972 to 1982. The recipient of the Greenberg Distinguished Lecturer Award is selected by the Biostatistics faculty based on the quality and public health impact of nominees’ research.