A1977
Title: High-dimensional error-in-variables survival analysis
Authors: Jinfeng Xu - City University of Hong Kong (Hong Kong) [presenting]
Abstract: High-dimensional survival data with measurement errors are prevalent in biomedical studies, where numerous clinical or genetic variables are collected for risk assessment. The presence of measurement errors in covariates substantially complicates parameter estimation and variable selection, often leading to challenging non-convex optimization problems. An error-in-variables additive hazards regression model for high-dimensional noisy survival data is proposed. By employing the nearest positive semi-definite matrix projection, a fast Lasso approach and its soft-thresholding variant are developed, both supported by strong theoretical guarantees. Under mild assumptions, model selection consistency, oracle inequalities, and asymptotic distributions are established for the proposed methods. Simulation studies and applications to two real datasets demonstrate the superior performance of these methods in handling high-dimensional data. Notably, they exhibit remarkable robustness even in scenarios with missing values, highlighting their practical utility in complex biomedical settings.
