SuperNOVA: Semi-Parametric Identification and Estimation of Interaction and Effect Modification in Mixed Exposures using Stochastic Interventions in R

Environmental epidemiology studies aim to understand the impact of mixed exposures on health outcomes while adjusting for covariates. However, traditional statistical methods make simplistic assumptions that may not be applicable to public policy decisions. Researchers are ultimately interested in answering causal questions, such as the impact of reducing toxic chemical exposures on adverse health outcomes like cancer. For example, in the case of PFAS, a class of chemicals measured simultaneously in blood samples, identifying the shifts that result in the greatest reduction in thyroid cancer rates can help more directly inform policy decisions on PFAS. In mixtures, nonlinear and non-additive relationships call for new statistical methods to estimate such modified exposure policies. To address these limitations, the open-source SuperNOVA package has been developed to use data-adaptive machine learning methods for identifying variable sets that have the most explanatory power on an outcome of interest. This package applies non-parametric definitions of interaction and effect modification to these variable sets in a mixed exposure, enabling researchers to explore modified treatment policies using stochastic interventions and answer causal questions. The SuperNOVA software implements the data-adaptive discovery of variable sets and estimation using optimal estimators for stochastic interventions described in our paper “Semi-Parametric Identification and Estimation of Interaction and Effect Modification in Mixed Exposures using Stochastic Interventions” (McCoy et al., 2023).