Flexible multivariate marginal models for analyzing multivariate longitudinal data, with applications in R.

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Traditional multivariate statistical models often require fitting separate parameters for covariate effects on multiple responses, which can be inefficient.
• There is a need for more flexible and efficient modeling strategies for analyzing multivariate longitudinal data.

Purpose of the Study:

• To propose a flexible modeling framework for multivariate marginal models to analyze multivariate longitudinal data.
• To demonstrate the model's capability to handle diverse response families including binomial, count, and continuous data.
• To introduce an R package, mmm2, for fitting these models.

Main Methods:

• Development of a multivariate marginal modeling framework.
• Application to the Kenya Morbidity dataset for illustration.
• Conducting a simulation study to assess parameter estimate accuracy.

Main Results:

• The proposed framework offers flexible model-building strategies for multivariate longitudinal data.
• The model successfully accommodates multiple response types (binomial, count, continuous).
• Simulation results indicate reliable parameter estimation.

Conclusions:

• The new modeling framework provides an efficient alternative to traditional methods for multivariate longitudinal data analysis.
• The mmm2 R package enables practical implementation of this flexible approach.
• This methodology is applicable to various fields dealing with complex, multi-response longitudinal data.