Semiparametric estimation in generalized linear mixed models with auxiliary covariates: a pairwise likelihood...

Area of Science:

• Biostatistics
• Statistical Modeling
• Biomedical Research

Background:

• Auxiliary covariates are common in biomedical research, with primary exposure measured in subgroups.
• Clustered data presents unique challenges for statistical analysis.
• Generalized linear mixed models (GLMMs) are widely used but can be sensitive to model assumptions.

Purpose of the Study:

• To propose a novel semiparametric estimation method for GLMMs with auxiliary covariate information in clustered data.
• To develop a robust inference procedure that treats error structure and random effects as nuisance parameters.
• To provide a method that is robust to misspecification and allows for covariate-random effect dependence.

Main Methods:

• A semiparametric estimation method utilizing a pairwise likelihood function.
• An estimating equation-based inference procedure.
• Treating error structure and random effects as nuisance parameters.

Main Results:

• The proposed method is robust against misspecification of error structure or random-effects distribution.
• The method allows for dependence between random effects and covariates.
• Asymptotic properties show estimators are consistent and asymptotically normal.
• Simulation studies demonstrate superior performance compared to validation set and existing methods.

Conclusions:

• The novel semiparametric method provides a robust and efficient approach for analyzing clustered data with auxiliary covariates.
• The method addresses limitations of existing techniques, offering improved performance and flexibility.
• Demonstrated utility in real-world biomedical data analysis.