home | research | cv | teaching | resources | extra-curricular
Research interestsInfectious diseases: Statistial methods for modeling spatial-temporal dynamics of disease processes and for estimating natural history parameters of disease.Missing or incomplete data: Likelihood approaches, imputation. Statistical computing: Reproducible research. Research projectsInfluenza epidemics: Using time-series data on influenza mortality in the US, we are working on adaptating existing models for this influenza time-series data that accurately captures the yearly seasonal trends in flu mortality across geographical regions in the US. [Joint work with Ron Brookmeyer]Incubation periods of respiratory viral diseases: Using likelihood-based methods to estimate incubation periods in the presence of double- or single-interval censored data, we are exploring the impacts of using data-reduction techniques when fitting parametric models to the data. [Joint work with Justin Lessler, Derek Cummings and Ron Brookmeyer] Red Cross blood supply: I work with a research team with students and faculty from the Biostatistics and Epidemiology departments here at Hopkins. Together with researchers from the Red Cross, we are investigating the impact of pandemic influenza (or other such possible public health crisis) on the blood supply in the US. [Joint work with Ming-Wen An, Stephen Crawford, Ron Brookmeyer, Ken Nelson, Tom Louis and the Red Cross team] Nutrition and frailty Using data from the Women's Health and Aging Study (WHAS), a cohort study of elderly women in the Baltimore area, we are investigating the relationship between various miconutrients and frailty. Those involved with WHAS have done some interesting research defining a phenotype of frailty based on measurable clinical symptoms. We are using this definition to find a micronutrient summary score that is predictive of frail status and then studying whether that score has broad utility when looking at other bad outcomes. [Joint work with Amy Mahoney, Carlos Weiss, Karen Bandeen-Roche, Richard Semba and Linda Fried] Genetic pathway analysis: Using data from a case-control study of carotid artery atherosclerosis disease (CAAD) based at the University of Washington, we explored pathway effects using a combination of random forest and logic regression methods. The non-parametric random forest methods were used to select a subset of covariates most predictive of CAAD status, and then we implemented Monte Carlo Logic Regression to identify specific relationships that were present in the dataset. [Joint work with Alex Nord et al.] Randomized interventions to turn out the vote: During the election cycle in 2006, classmate Jessica Myers and I helped two environmental non-profit groups design studies to test whether various get-out-the-vote tactics are indeed turning out the vote. To assist the groups in deciding how large to make the test and control groups, we designed a simulation to estimate the power of detecting an increase in voter turnout (if it exists) under different circumstances. We also performed the follow-up analysis. |
615 N. Wolfe St. Baltimore, MD 21205 |
office E3035 410.502.3364 |