My methods work primarily focuses on multi-pollutant modeling, non-linear interaction selection in the presence of many environmental contaminants, and statistical inference when exposure data are subject to multiple detection limits. I have a number of applied collaborations in environmental statistics ranging from amyotrophic lateral sclerosis, racial disparities in telomere length, and gestational duration in pregnant women.
My main research interest is developing methods for determining noteworthy associations from genetic studies. Accurate assessment of potential signals from the exploratory stage is important for minimizing false-positive findings for follow-up analysis.
I am passionate in utilizing data integration techniques to improve the precision and efficiency of disease risk prediction. As my PhD thesis, I am developing efficient statistical tools to incorporate external information from big health science data into internal studies.
Currently, I am working on quantifying gene-environment interactions on the prevalence of cancer. I would like to see which measures of risk associations are the most informative on providing the most accurate quantification of the relationship between gene-environment interaction and the prevalence of disease.
My current research involves the use of polygenic risk scores, which are essentially weighted sums of effect allele counts, to predict a patient's genetic risk for certain diseases. In particular, I am interested in comparing and assessing different methods of generating polygenic risk scores in terms of power and predictive accuracy, with the goal of improving individual disease prediction. I am currently supported by the Biostatistics Training in Cancer Research grant.
I am interested in causal inference methods and their applications to cancer epidemiology. My work on analysis of subpopulation causal effects helps bridge the gap between traditional population causal inference and individualized inference of cancer treatment.
I currently work on developing high-dimensional Bayesian mediation analysis methods with application to omics data. I am also interested in developing analytical approaches to extract information from large-scale data, with the hope to advance our understanding of the molecular mechanism of human diseases.
Yaoxuan (Vivian) Xia
I'm interested in survival analysis and statistical genetics. I have worked on using segregation analysis to predict the risk of cancer in mutation carriers, and using bayesian models to evaluate differential diagnoses and multiplex gene testing.
My current research focuses on causal inference for censored data. I'm working on a number of projects including evaluating the effect of treatments for metastatic prostate cancer using insurance claims data.
I am interested in the environmental risk score and already published one paper titled "Construction of environmental risk score beyond standard linear models using machine learning methods: application to metal mixtures, oxidative stress and cardiovascular disease in NHANES" in Environmental Health. Currently, I am working on the joint model of latent class analysis.