Research

Due in part to my experience working at and working with the CDC, much of my work in spatial statistics and spatial epidemiology pertains to the analysis of county-level health outcome data.  From a methodological standpoint, my research aims to develop statistical models that leverage complex dependencies between data points, whether it be spatial and temporal structures or relationships between demographic strata (e.g., age groups, gender, and race/ethnicity).

My work in this area is the focus of my recently funded NIH/NHLBI R01.

When analyzing geographically referenced health outcome data stratified by various demographic factors, you inevitably encounter an abundance of small counts.  Unfortunately, publicly releasing these small counts can come with a high risk of disclosure -- e.g., if an individual is unique based on a small number of attributes (e.g., age, race, gender, county of residence, and cause of death), they may be identifiable in otherwise de-identified data, thereby risking the disclosure of other attributes about that individual that were not intended to be publicly disclosed.  My research in disclosure limitation / data confidentiality pertains to the application of methods designed for the analysis of spatial data for the purpose of generating synthetic data that provide  the same or similar inference as the original, potentially sensitive data but that can be released for public-use without the associated risks of disclosure.

My work in this area is the focus of my recently funded NSF CAREER award.

A frequent collaborator of mine, Dr. Tran Huynh, is a certified industrial hygienist with a background in occupational exposure assessment.  Because occupational exposure data often consists of limited data and measurements below detectable limits, our work together has focused on the design and implementation of Bayesian methods that leverage historical data to improve the precision of workers' exposures.