Flexible inference of optimal individualized treatment strategy in covariate adjusted randomization with multiple covariates

To maximize clinical benefit, clinicians routinely tailor treatment to the individual characteristics of each patient, where individualized rules are needed and significant research interest statisticians. In covariate-adjusted randomization trial with many covariates, we model effect an unspecified function a single index covariates leave baseline response completely arbitrary. We devise class estimators consistently estimate its associated while bypassing estimation response, which is subject curse dimensionality. further develop inference tools identify predictive isolate effective region. The usefulness methods demonstrated in both simulations data example.