Biostatistics/Epidemiology Training Grants in AIDS



Current Fellows (2013-2014)

Sarah Anoke is a second-year student in the Biostatistics doctoral program; her dissertation directors are Drs. Cory Zigler and Giovanni Parmigiani. Her primary research interest is the application of causal inference to infectious disease epidemiology. During her first summer in the program, she worked under of the direction of Dr. Miguel Hernan to reproduce a simulation study comparing dynamic HIV antiretroviral treatment regimes that compared mortality rates after starting ART at different CD4 cell count levels. Observational data was used to estimate covariate and outcome distributions, and to establish baseline values. Subsequent covariate values were simulated, and mortality rates estimated using causal methods. Sarah is also interested in survey methodologies as they relate to health system monitoring and evaluation in Africa. She is currently working on a report that details a formal statistical comparison of health metrics (e.g. HIV testing prevalence, antenatal care attendance, child nutrition) from two well-known surveys recently conducted in Uganda.

Katherine Evans is a second-year student in the Biostatistics doctoral program whose research interests are the application of causal inference to infectious disease epidemiology. Her dissertation advisor is Dr. Eric Tchetgen Tchetgen.

Emily Smith is a second year student in the Global Health and Population doctoral program (academic advisor: Professor Wafaie Fawzi). Emily is involved in two research projects. She is designing a protocol to collect HIV status information for women and infant that are currently enrolled in an individually randomized, double blind, placebo controlled trial in Tanzania that is designed to assess the efficacy of neonatal vitamin A supplementation given within the first three days of life in reducing mortality in the period from supplementation to 6 months of age. She is also working on an analysis of baseline ferritin status and HIV progression among participants in a randomized double blinded placebo controlled trial to assess the efficacy of multivitamin supplementation in HIV infected adults initiating antiretroviral therapy in Uganda in achieving immune reconstitution, weight gain, and improved quality of life.

Yared Gurmu is a third year student in the Biostatistics doctoral program, and his dissertation advisor is Dr. Victor DeGruttola. His research interests include the application of statistical methodologies to address problems that arise in the design and analysis of clinical trials for HIV/AIDS and other infectious diseases. Last year, he provided data analysis support to Dr. Stella Safo on a research project that examined the effectiveness of "Peer to Peer" youth HIV/AIDS education in two Dorchester High Schools in Boston, MA. Currently, Mr. Gurmu has been reviewing and developing methods of nonparametric survival analysis; his goal is to develop a methodology that permits accurate estimation of joint distribution of start time and duration of relationships and apply this methodology to HIV surveillance data. This methodology will also be extended to examine the impact of age, gender and treatment status on the distribution of start times and duration of relationships.

Denis Agniel is a fourth-year student in the Biostatistics doctoral program. His dissertation advisor is Professor Tianxi Cai. Denis's research focuses on developing methods to work with high dimensional data. For example, he is working with Professor Cai to identify so-called "master regulators," or predictors (often genetic) that are associated with multiple related outcomes using a wide class of models including semi-parametric regression models. They used this to try to identify a common genetic basis for autoimmune disease and for drug resistance to entire classes of drugs in HIV. He is also working on a project to identify SNPs that affect CD4 count over time, using functional principal components analysis. The functional PCA approach allows him to detect more complex effects than would have been possible in conventional longitudinal analyses. Denis won the David P. Byar Young Investigator Award for this work and presented "Identifying master regulators in semi-parametric regression models" at the Joint Statistical Meeting in Quebec in August of 2013. Denis is currently working on a study of host control of HIV disease progression in two cohorts of HIV-infected but treatment-naive subjects from Botswana. HIV disease progression was measured as longitudinal measurements of CD4 cell count and HIV RNA (viral load). These measurements were collected as a part of a collaboration fostered by the Harvard-Botswana partnership. A large subset of the individuals have been genotyped for this the study, and Denis has worked to develop novel statistical methods to identify locations in the genome which affect disease progression; some of the identified regions may tell researchers why some people control the virus well in the absence of antiretroviral treatment.

Caleb Miles is a fourth-year student in the Biostatistics doctoral program, and currently working on his dissertation under the direction of Dr. Eric Tchetgen Tchetgen. Caleb's research focuses on the development of identification and estimation of a causal contrast in the field of mediation analysis. This contrast is of particular interest in settings where there may be both confounders of the effect of the exposure on the outcome as well as confounders of the effect of the mediator on the outcome that is affected by the exposure. One such setting is where there is interest in how adherence to a drug regimen mediates the effect of prescription to an anti-retroviral therapy on the outcome, virologic suppression. In addition to confounding variables on the effect of the ART on virologic suppression, it is also important to account for potential toxicity due to exposure to the ART, which could affect both adherence and the outcome. The causal contrast Caleb is studying captures specifically the effect of the exposure on the outcome exclusively through the mediator and not through the post-exposure confounders. He is developing identification assumptions for this contrast, and using a semiparametric approach to produce an estimator that is both efficient and multiply robust.

Elizabeth Smoot is a fourth-year student in the Biostatistics doctoral program, currently working on her dissertation under the direction of Dr. Sebastien Haneuse. Ms. Smoot is working on building on the hybrid design for combining ecological and case-control proposed by Haneuse and Wakefield. A main obstacle to the applicability of the hybrid design is its large computational burden. A potential solution to this is to use an alternative representation of the hybrid likelihood combined with a series of approximations to drastically reduce the computational burden of the hybrid design while maintaining the operating characteristics of the maximum likelihood point estimates. A second potential solution is to use calibration methods in conjunction with influence functions. These methods are motivated by and useful in resource-limited settings, which include monitoring of HIV/AIDS patient outcomes in national antiretroviral treatment programs for resource-poor countries such as Malawi.

Natalie Exner (dissertation advisor: Marcello Pagano)is a fifth-year student in the Biostatistics doctoral program. There are three components of her thesis. The first project relates to HIV incidence estimation using cross-sectional viral genetic diversity, and the second project assesses methods for confidence interval estimation for proportions in survey settings. For her third project, Natalie has been consulting for the World Health Organization as they update their protocols for HIV drug resistance surveillance in low- and middle-income countries. The goal has been to create a general survey design that works well for a variety of countries. She has applied survey design methodology as well as developed alternative approaches for survey design calculations to decrease costs while maintaining statistical efficiency. This year, Natalie presented at regional workshops in Kunming, China, and Brasilia, Brazil, to meet with country program managers to assess feasibility. She also attended a Steering Group meeting in Geneva, Switzerland, to discuss the updated protocols. Natalie has received several awards for her work, including two from the American Statistical Association. After graduating in May 2014, Natalie will work for the World Health Organization.