Estimating individualized treatment rules for treatments with hierarchical structure

Precision medicine is an increasingly important area of research. Due to the heterogeneity individual characteristics, patients may respond differently treatments. One most goals for precision develop individualized treatment rules (ITRs) involving patients’ characteristics directly. As interesting topic in clinical research, many statistical methods have been developed recent years find optimal ITRs. For binary treatments, outcome weighted learning (OWL) was proposed a decision function patient maximizing expected outcome. Treatments with hierarchical structure are commonly seen practice. In scenarios, how estimate ITRs still unclear. We propose new framework named outcome-weighted angle-based (HOAL) treatments structure. Statistical properties including Fisher consistency and convergence rates method presented. Simulations application type 2 diabetes study under linear nonlinear show highly competitive performance our procedure both numerical accuracy computational efficiency.