A0632
Title: Gene expression analysis with SIMEX-trans: Two-phase transfer learning with covariates subject to measurement error
Authors: Kaida Cai - Southeast University (China) [presenting]
Abstract: The focus is on high-dimensional linear regression in the context of transfer learning, particularly when covariates are subject to measurement errors. A two-phase high-dimensional transfer learning method is proposed using a simulation-extrapolation procedure to improve the performance of target data analysis by leveraging information from auxiliary data sets. Furthermore, the method can adjust estimates to account for the impact of measurement errors when covariate measurement errors are present in both target and auxiliary data sets. In the simulation studies, the method is compared with others that either disregard information from auxiliary data sets or ignore the effect of measurement errors. The results of estimation errors demonstrate that our method outperforms other methods across various scenarios involving different settings of auxiliary data sets and magnitudes of measurement error. The proposed method is applied to gene expression data analysis, showing that it improves gene expression prediction performance in a target tissue by adjusting the effect of measurement errors and integrating information from multiple tissues as auxiliary data sets.
