Inference for Treatment Regime Models in Personalized Medicine

摘要: In medical practice, when more than one treatment option is viable, there little systematic use of individual patient characteristics to estimate which most likely result in a better outcome for the patient. We introduce new framework using statistical models personalized medicine. Our exploits (1) data from randomized comparative trial, and (2) regression model constructed domain knowledge no requirement correct specification. "improvement" measure summarizing extent model's allocations improve future subject outcomes on average compared business-as-usual allocation approach. Procedures are provided estimating this as well asymptotically valid confidence intervals. One may also test null scenario hypothesized not useful demonstrate our method's promise simulated experiment testing treatments depression. An open-source software implementation procedures available within R package "Personalized Treatment Evaluator" currently CRAN PTE.