Liangyuan Hu, PhD (she/her/hers)
Liangyuan Hu, Ph.D., is an associate professor in the Department of Biostatistics and Epidemiology at the Rutgers School of Public Health. Prior to joining Rutgers University, Dr. Hu was an assistant professor of biostatistics in the Department of Population Health Science and Policy at Mount Sinai School of Medicine. Dr. Hu received her doctoral degree in biostatistics from Brown University.
Dr. Hu's research interests include statistical methods for causal inference, missing data and Bayesian inference, with applications in cancer, Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome (HIV/AIDS) and cardiovascular diseases.
Dr. Hu and colleagues developed a continuous-time marginal structural model and statistical methods to estimate the causal effect of treatment initiation timing on mortality from electronic health records data. The methods were applied to find out about the optimal timing of antiretroviral therapy initiation for patients who are presented with both Human Immunodeficiency Virus (HIV) and Tuberculosis (TB). The models offered detailed understanding of the effects of treatment timing versus treatment duration on mortality and can be used to emulate clinical trials to compare with findings from randomized, controlled trials. This work has won the Outstanding Statistical Application Award from the American Statistical Association in 2019. Funded by Patient-Centered Outcomes Research Institute (PCORI) and the National Institutes for Health (NIH), Dr. Hu is leading a team to develop statistical methods to improve comparative effectiveness research in the context of multiple treatments and longitudinal treatments. Our team has developed several Bayesian machine learning based approaches that allow for drawing more accurate causal inferences about treatment effects using complex health datasets and assessing how sensitive the causal conclusions are to varying degrees of unmeasured confounding. We have created three open-source software packages, CIMTx, SAMTx and riAFTBART, available on CRAN, for fellow researchers to implement our methods.