A1199
Title: Synthesizing auxiliary subgroup restricted mean survival information to obtain efficient estimation for Cox model
Authors: Jo-Ying Hung - National Cheng Kung University (Taiwan) [presenting]
Pei-Fang Su - National Cheng Kung University (Taiwan)
Abstract: Restricted mean survival time (RMST) has gained attention in the medical research field because it is model-free and easy to interpret. Due to the increasing accessibility of data sources, it is of interest to make use of published external information, such as RMST, to increase the efficiency of estimators in individual-level studies. In this research, a double empirical likelihood method is proposed that incorporates auxiliary RMST into the estimation of the Cox proportional hazard model. It is proved that the proposed estimator asymptotically follows a multivariate normal distribution and is asymptotically more efficient than the classical partial likelihood (PL) estimator. Simulation studies show that the proposed method yields more efficient estimators compared to the PL estimators. A diabetes dataset is used to demonstrate the proposed method.