SNP_NLMM: A SAS Macro to Implement a Flexible Random Effects Density for Generalized Linear and Nonlinear Mixed...

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Generalized linear and nonlinear mixed models (GLMMs and NLMMs) are standard for longitudinal data.
• A key assumption is Gaussian random effects, which can be biologically unrealistic.
• Misspecifying random effects distributions leads to biased and inefficient parameter estimates.

Purpose of the Study:

• To develop a computational method for fitting GLMMs and NLMMs with flexible random effects densities.
• To overcome limitations of previous methods that required Gaussian random effects.
• To provide a practical tool for analyzing complex longitudinal data where random effects may not be Gaussian.

Main Methods:

• Development of a SAS macro, SNP_NLMM.
• Implementation of a seminonparametric (SNP) density formulation for random effects.
• Accommodating flexible response distributions and nonlinear mean trajectories.

Main Results:

• The SNP_NLMM macro successfully fits GLMMs and NLMMs with flexible random effects densities.
• Demonstrated application on a GLMM for toenail infection disease progression.
• Showcased utility on a NLMM for intravenous drug concentration data.

Conclusions:

• The SNP_NLMM macro offers a computationally feasible approach for flexible random effects modeling.
• This method enhances the accuracy and reliability of analyses for non-Gaussian or nonlinear mixed models.
• Facilitates more robust statistical inference in longitudinal and clustered data analysis.