A0339
Title: Estimation of treatment effects based on robust sparse reduced-rank regression
Authors: Ryoma Hieda - Doshisha University (Japan) [presenting]
Shintaro Yuki - Doshisha University (Japan)
Kensuke Tanioka - Doshisha University (Japan)
Hiroshi Yadohisa - Doshisha University (Japan)
Abstract: In clinical trials, we are interested in the estimation of heterogeneous treatment effects (HTE) to develop strategies for personalized medicine. We focus on the modified covariate method (MCM) as one of the methods to estimate the HTE. The model of MCM includes the term of interaction between the treatment and covariates without the main effects and is formulated for a single outcome. However, in clinical trials, there can be interest in multiple outcomes, such as primary and secondary endpoints. Therefore, we extended MCM to the case of multiple outcomes. In addition, we observed that data from clinical trials could include outliers and highly correlated outcomes. In such cases, the HTE cannot be properly estimated. Hence, we propose a method to estimate HTE using MCM in a framework of Robust sparse reduced-rank regression. The proposed method improves the accuracy of estimating HTE because it can deal with highly correlated outcomes by setting rank constraints on the regression coefficient matrix for treatment effects and removing the effects of outliers. We demonstrate the effectiveness of the proposed method based on simulation and real data examples.
