Robust Sample Weighting to Facilitate Individualized Treatment Rule Learning for a Target Population

Learning individualized treatment rules (ITRs) is an important topic in precision medicine. Current literature mainly focuses on deriving ITRs from a single source population. We consider the observational data setting when population differs target of interest. Compared with causal generalization for average effect which scalar quantity, ITR poses new challenges due to need model and generalize based prespecified class functions may not contain unrestricted true optimal ITR. The aim this paper develop weighting framework mitigate impact such misspecification thus facilitate generalizability Our method seeks covariate balance over non-parametric function characterized by reproducing kernel Hilbert space can improve many learning methods that rely weights. show proposed encompasses importance weights overlap as two extreme cases, allowing better bias-variance trade-off between. Numerical examples demonstrate use our greatly estimation compared other methods.