syhyunpark/csim: Constrained single index models (CSIM)

A major challenge in estimating individualized treatment selection rules from an RCT dataset lies in detecting relatively small treatment effect modification-related variabilities (i.e., the treatment-by-covariates interaction effects on the treatment outcome) against a relatively large non-treatment-related variabilities (i.e., main effects of covariates on the treatment outcome). The class of constrained single index models (CSIM) is a novel single-index model specifically designed to estimate a component associated with the treatment effect modification-related variabilities, while allowing for a nonlinear interaction effect between the treatment and a potentially large number of pretreatment covariates. CSIM provides a flexible regression approach to developing individualized treatment selection rules based on patients' data measured at baseline. For details, see “A Constrained single index model for estimating interactions between a treatment and covariates” (Park, H., Petkova, E., Tarpey, T., Ogden, R.T., 2019). The main function of this package is csim().

install.packages("remotes")
remotes::install_github("syhyunpark/csim")