| With the progress of modern science and medical technology,human’s demand for medical treatment is increasing year by year,which promotes the transformation from traditional ”onesizefitsall” treatment to personalized medicine.Personalized medicine,also known as precision medicine,aims to use cuttingedge medical technology to tailor treatments to individual differences.In biometrics,an important direction of precision medicine using causal inference theory is the estimation of the optimal individualized treatment rule.The optimal individualized treatment rule is to recommend different treatment regimens according to the patient’s characteristics,so as to maximize the overall average expected clinical outcome of the patients.In this paper,empirical likelihood is used to improve the inverse propensity score weighted estimation and then we proposed multiple robust estimates for the optimal individualized treatment rule.We first describe in detail the basic framework for estimating the optimal individualized treatment rule,noting that since most approaches rely on the specification of propensity score models,robustness of estimates is critical,and then we introduce augmented inverse propensity score weighted estimation with doubly robustness.Then,using the empirical likelihood,we propose multiple robust estimators for the propensity score model under the missing at random,and prove the consistency and asymptotic normality.Finally,the multiple robust estimators are applied to the numerical simulation and ACTG175 data and compared with the results of other estimators.The simulation results show that compared with IPSW,the proposed multiple robust estimation is easier to achieve better estimation effect. |
