My research interests focus on two overarching areas. First, I am interested in investigating the comparative effectiveness of treatments for urologic cancers. Although randomized clinical trials are the preferred study design to evaluate the comparative effectiveness of interventions, there are very few trials comparing surgical treatments within urologic oncology, in part due to the difficulty of conducting surgical trials. To address such fundamental evidence gaps, I am interested in the application of two novel observational research methods when clinical trial evidence is limited: emulation of target clinical trials using observational datasets, and transportation of inferences from competed clinical trials to “real-world” patient populations.
My other research interest is the development of deep learning methods to improve the diagnosis and risk-stratification of urologic cancers. Together with collaborators from Brown University, we have developed deep learning methods to improve the histopathologic diagnosis and Gleason grading of prostate cancer biopsy specimens. Additional work is ongoing to extend these efforts to other prediction tasks.