Low-dimensional confounder adjustment and high-dimensional penalized estimation for survival analysis

Area of Science:

• Biomedical research
• Genomics
• Biostatistics

Background:

• High-throughput gene expression profiling is prevalent in biomedical research.
• Current methods for etiology studies often use marginal screening and penalized regression, but may not adequately account for confounders.
• Confounding factors like clinical risks and environmental exposures can impact gene expression analysis.

Purpose of the Study:

• To propose covariate-adjusted screening and variable selection procedures for etiology studies using gene expression data.
• To improve the accuracy of regression coefficient estimation by adjusting for low-dimensional confounder effects.
• To develop methods that properly account for confounders in the context of the accelerated failure time model.

Main Methods:

• Development of covariate-adjusted screening and variable selection procedures.
• Application of penalized regression for model building while adjusting for confounders.
• Theoretical establishment of asymptotic properties for the proposed methods.
• Simulation studies to evaluate finite sample performance.

Main Results:

• The proposed methods provide more accurate estimation of regression coefficients by adjusting for confounders.
• Asymptotic properties of the new procedures have been theoretically established.
• Simulation studies demonstrate the effectiveness of the methods in finite samples.

Conclusions:

• The developed covariate-adjusted methods enhance the analysis of gene expression data in etiology studies.
• Proper adjustment for confounders leads to more meaningful and reliable results in real-world data analysis.
• This approach offers a more robust framework for high-throughput profiling in biomedical research.