Predoctoral Fellows (2024-2025)
Stephanie Wu
Academic Advisor: Briana J Stephenson / Michael D Hughes
Jemar is a fifth year PhD student in the Department of Biostatistics. Stephanie’s research interests were motivated by a collaborative research project mentored by Dr Hughes in which she considered outcome-dependent case-cohort designs nested within a tuberculosis/HIV cluster randomized trial being undertaken by the ACTG. With program faculty member, Dr Hedt-Gauthier, Stephanie also worked on statistical methods and tools for World Health Organization protocols on laboratory- based, clinic-based, and sentinel surveys that allow countries to assess levels of acquired HIV drug resistance using remnant viral load specimens. These two projects and coursework led to a broader interest in high dimensional clustered data of the type that arises in nutritional studies. Her thesis research therefore led to a focus on statistical methods to address clustering of high dimensional exposure variable and their association with outcomes, with applications to nutritional epidemiology and infectious diseases. Her first thesis paper presents a new statistical method that uses a supervised Bayesian clustering framework to account for complex survey designs when analyzing high-dimensional dietary data and their association with disease outcomes.
Maxwell Wang
Academic Advisor: Jukka-Pekka Onnela
Christina is a fourth year PhD student in the Department of Biostatistics. Max’s research interest is in the use of contact network information for applications in clinical trials and epidemic response. Currently, he is exploring methods for simulation-guided Bayesian inference for epidemics. In particular, he is considering a method dubbed the MDN-ABC, which allows for the flexible use of Approximate Bayesian Computation without the need to specify summary statistics or make any parametric assumptions about the form of the posterior distributions. Max hopes to expand this framework to handle situations where epidemics are only partially observed (event times and contact networks are not fully known). In one project he focused on the subject of clustered randomized trials (CRTs). In situations where a CRT is run in conjunction with a spreading contagion, he sought to show via simulation that features of a contact network (e.g., distance to nearest affected neighbor) can be used in generalized estimating equation (GEE) augmentation. This augmentation strategy can lead to both improved efficiency and power in effect estimation.
Daniel Xu
Academic Advisor: Rajarshi Mukherjee
Daniel is a fourth year PhD student in the Department of Biostatistics. Daniel is interested in developing methods for causal inference in the semi-supervised setting. Daniel is currently working on two projects. In the first project, he is seeking to understand the performance of a novel estimator of the average treatment effect in the semi-supervised setting where the outcome is missing at random, as well as apply it to a real-world dataset. This project is nearing completion. In the second project, he is seeking to understand the performance of estimators of quantities relevant to causal mediation in the semi-supervised setting, and then again to apply the resulting estimator to a real-world dataset. Simulation studies and theoretical results have already been conducted for both projects.
Keith M Barnatchez
Academic Advisor: Rachel Nethery
Keith is a third year PhD student in the Department of Biostatistics. Keith’s research interest is in methods for causal inference when exposure or confounding variables are subject to measurement error. The initial goals are 1) to review the concepts of causal inference and measurement error in a unifying framework, and 2) to provide a comprehensive review of the existing methods for addressing measurement error in causal inference, with simulation studies to assess their relative performances. Over the past year he has developed a simulation study that compares the performance of several causal effect estimators that address measurement error (either in the exposure variable itself or in a confounder). He is also working on a causal inference method for flexibly addressing numerous types of measurement error, with the ability to accommodate varying study designs. He has proved various theoretical properties of the method and developed simulation studies to compare the control variates estimator to current gold-standard methods under various plausible conditions. Currently, Keith is working with investigators at Vanderbilt University to obtain data for an application that focuses on individuals living with HIV receiving care from the Vanderbilt Comprehensive Care Clinic.
Emma G Krenshaw
Academic Advisor: Jukka-Pekka Onnela
Emma is a third year PhD student in the Department of Biostatistics. Emma’s research interest is the use of mechanistic network models for infectious diseases and statistical methodology for these networks. Her current work focuses on expanding methods to model transitivity in networks to better capture local network structure without losing the ability to model higher-order structure as well. In practice, this would allow her to better model complex network structure that determines the spread of epidemics on networks. Emma has designed and implemented a complex agent-based mechanistic network (CAN) model in Python to simulate the 2022 mpox outbreak among men who have sex with men. This work is motivated by previous CAN models for HIV in this population. Using this model, she can simulate the effect of various interventions and behavioral changes on the progress of an outbreak of mpox, and can also assess the influence of different network characteristics on the simulated epidemic, such as frequency of partner concurrency. Emma has also been interested in applications of network science to cluster randomized trials in the context of infectious disease. She worked on a network science research project with Dr. Jukka-Pekka Onnela that focused on two chapters in Mark Newson’s “Networks” textbook related to epidemics and percolation. She also worked on a more applications-based project looking at the application of network science to infectious disease transmission, focusing on studies of HIV transmission.
Lee J Ding
Academic Advisor: Michael D Hughes
Lee is a second year PhD student in the Department of Biostatistics. Lee’s current research is related to designing adaptive cluster-randomized trials (CRTs), with application to HIV-related research. A motivating example is the PHOENIx CRT being undertaken by the ACTG clinical trials network which is evaluating treatment to prevent development of multi-drug resistant tuberculosis (MDR-TB) among high-risk household contacts, including people living with HIV, of a patient with MDR-TB. He is currently developing methods for estimating the maximum and expected sample sizes when designing group sequential cluster-randomized trials that can stop early for efficacy and/or futility. He is working on a simulation study for the project and plans to begin assembling the manuscript afterwards.
James Dominic DiSanto
Academic Advisor: Rachel Nethery
Dominic is a second year PhD student in the Department of Biostatistics. Dominic’s research interests are focused on using electronic health records to address important clinical questions. He was funded on a university scholarship during his first year in the PhD program. As a research project during this time, he led an effort in developing semi-supervised causal modeling methods for generating real world evidence on comparative effectiveness of two drugs for treating multiple sclerosis using registry linked longitudinal EHR data. The clinical outcome of interest was a disability score which is only ascertained in patients belonging to the registry while a majority of the study population only had EHR data. Dominic developed a robust imputation strategy to impute the missing outcome for patients in the EHR, while correcting for potential model mis-specification, confounding, as well as bias due to heterogeneity in
patient characteristics between the registry and EHR cohorts. The algorithm development and interpretation of results were refined over time by working together with an interdisciplinary team with expertise in clinical medicine, biomedical informatics, and biostatistics. These proposed methods are broadly applicable to EHR- based clinical research and he is shifting focus to HIV-related comparative effectiveness problems.
Nicolas W. Birk
Academic Advisor: Rui Wang
Nick has completed a semester of coursework. He is starting a research project in the next semester focused on stratified medicine/personalized treatment for ordinal outcomes, particularly risk/benefit composite outcomes with applications in HIV co-infection and co-morbidity clinical trials.
Sarah J. Boese
Academic Advisor: Paige Williams
Sarah is a first year PhD student in the Department of Biostatistics. Sarah has completed a semester of coursework. She is developing ideas for a research project starting in the next semester based on her interests in health disparities among people living with HIV, spatio- temporal statistics and causal inference.
Christian C. Testa
Academic Advisor: Nima Hejazi
Christian is a first year PhD student in the Department of Biostatistics. Christian has just completed the first semester of coursework in the PhD program. Based on prior research experience ad interaction with faculty at HSPH, he has research interests in measuring algorithmic bias using tools from mediation analysis in causal inference; and assessing the impacts of uncertainty in population denominators on epidemiological models, as well as designing new models to synthesize novel sources for population estimates (satellite imagery, mobility data, crisis population data) which might be useful, when combined with case data from other sources, in assessing incidence and prevalence of infectious diseases.