CMStatistics 2015: Start Registration
View Submission - CMStatistics
B0223
Title: Concordance-assisted learning for estimating optimal individualized treatment regimes Authors:  Wenbin Lu - North Carolina State University (United States) [presenting]
Abstract: A new concordance-assisted learning (CAL) is presented for estimating optimal individualized treatment regimes. First, we will introduce a type of concordance function for prescribing treatment and propose a robust rank regression method for estimating the concordance function. Then, we will discuss the proposed CAL methods for estimating optimal treatment regimes that maximize the concordance function, named prescriptive index, and for searching the optimal threshold. Moreover, we will discuss the convergence rates and asymptotic distributions of the proposed estimators for parameters in the prescriptive index and the optimal threshold. Finally, we will present some simulations and an application to an AIDS data to illustrate the practical use and effectiveness of the proposed methodology.