I am interested in the application of causal inference methods in observational studies, especially in the setting of longitudinal data, where time-varying exposures, time-dependent confounding, and selection bias are often present. In particular, I am interested in the application of the variety of modern analytical techniques for estimating causal effects as well as the concepts such as direct and indirect effect decomposition (causal mediation analysis), causal interactions and sufficient causes, and causal interference and spill-over effects. These analytical techniques can be applied to broad areas of health sciences research, but I am particularly interested in the application of causal inference for comparative effectiveness research and implementation science. The methods are very useful for the development and assessment of strategies to improve population health using non-experimental data.

I am also interested in the application of Bayesian methods for evaluation of diagnostic tests and prevalence estimation in the absence of a perfect reference standard. These methods have broad applications in diagnosis and screening, and can be used to estimate true prevalence using imperfect tests.