Meeting Calendar

Upcoming events

11/6/2026

Friday 1:30-2:45pm, 

Place: Genome Cafe, Room E3609

Light lunch 1:10-1:30pm

Zoom Link: https://jh.zoom.us/my/wange3614?omn=92414805340

Speaker: Rajeshwari Sundaram, Senior Investigator, Biostatistics & Bioinformatics Branch, Eunice Kennedy Shriver National Institutes of Child Health and Human Development, NIH

Title: From Complex Mixtures to Clear Inference: Statistical Methods for Environmental Health

Abstract: Environmental mixture studies present substantial statistical challenges because exposures are highly correlated, may have nonlinear effects, and are frequently measured below the limit of detection (LOD) or maybe lipophilic, while health outcomes may involve longitudinal processes and survival times. We have developed an integrated framework for variable selection and inference that jointly addresses these features within complex environmental mixture analyses. The framework combines a joint model for skewed longitudinal and discrete survival outcomes with shared frailty. We developed Discnet for selecting correlated exposures in discrete time-to-event analyses and conducted principled evaluation of LOD accommodation strategies for downstream mixture methods. These approaches enable selection of important exposures while accounting for dependence between outcomes, exposure correlation, censoring, nonlinear relationships, and measurements below the LOD.

Simulation studies demonstrate reliable variable selection and estimation for the joint model, stable false-positive and false-negative rates, and meaningful differences in mixture-analysis performance across LOD strategies. Applications to the LIFE Study and NHANES illustrate the utility of the framework for identifying important environmental exposures across reproductive and other health outcomes. Together, these methods advance statistical methodology for complex mixture studies by providing a coherent approach to exposure selection and inference across longitudinal, survival, and censored exposure data.

3/5/2027

Speaker: Fan Li, Professor, Dept of Statistical Science, Duke University

Tentative Title: Sample size and power calculation for causal inference with time-to-event outcomes

3/26/2027

Speaker: Charlie Wolock, Assistant Professor, Department of Biostatistics and Computational Biology, University of Rochester