Innovative Methods for Treatment Effect Heterogeneity & Calibration

Abstract

Treatment effect heterogeneity, which pertains to the identification of subgroups that have varyingresponses to the same treatment, is an increasingly important topic for personalized medicine and for validating the robustness of a novel treatment. Virtual Twins is an interpretable and flexible framework for the evaluation of treatment effect heterogeneity that has been highly cited. However, there are many modeling choices within Virtual Twins that can impact its operating characteristics, which have not been explored since its publication. Our first project presents a detailed analysis of Virtual Twin's performance under a variety of modeling choices and realistic problem settings, with an application to a randomized trial of very low nicotine content cigarettes. Often in public health, we are interested in the treatment effect of an intervention on a population that is systemically different from the experimental population the intervention was originally evaluated in. When treatment effect heterogeneity is present in a randomized controlled trial, generalizing the treatment effect from this experimental population to a target population of interest is a complex problem; it requires the characterization of both the treatment effect heterogeneity and the baseline covariate mismatch between the two populations. Despite the importance of this problem, the literature for variable selection in this context is limited. Our second and third projects present three novel methods, GLAVeS, BAC and WR, for variable selection in the context of treatment effect generalization, with an application to generalize the treatment effect of very low nicotine content cigarettes to the overall U.S. smoking population.