Several of my postdoctoral research projects focus on dynamic treatment regimes for longitudinal data with applications in HIV, including work with researchers in the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) group. Additional project areas include machine learning methods for risk score prediction in the epidemiology of aging and continued work in the area of case-control studies.
My dissertation developed targeted maximum likelihood estimators (TMLEs) for case-control designs. The data-generating experiment in case-control study designs involves an additional complexity called biased sampling. Case-control study designs are frequently used in public health and medical research to assess potential risk factors for disease. These designs are particularly attractive to investigators researching rare disease, as they are able to sample known cases of disease, vs. following a large number of subjects and waiting for disease onset in a relatively small number of individuals.
Our case-control-weighted TMLE relies on external knowledge of the true prevalence probability, or a reasonable estimate of this probability, to eliminate the bias of the sampling design in case-control weights. TMLEs are double robust efficient loss-based substitution estimators. The first step in the procedure is an estimate of the conditional expectation of the outcome given the exposure and baseline covariates. This estimate is updated using the probability of the exposure given covariates in a step targeted towards making an optimal bias-variance tradeoff for the parameter of interest.
During my Ph.D., I also worked on prediction methods, applied projects in biology, genetics, and epidemiology, and didactic presentations of causal inference methods.