Estimation of the optimal treatment regimes with multiple treatments under proportional hazards model

We propose a novel proportional hazards model that include an unknown baseline covariate effect and the interaction between multiple treatments and covariates on censored survival data. Doubly robust estimating equations constructed by utilizing the A-learning methodology and time-dependent propensity score. The asymptotic properties of the proposed estimators are established under the correct specification of either the baseline effect model or the propensity score model. Extensive simulation studies, along with an application, demonstrate the efficacy of the proposed approach.