A transition copula model for analyzing multivariate longitudinal data with missing responses.

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Multivariate longitudinal studies involve repeated measurements of multiple outcomes over time.
• These studies present challenges in modeling both cross-sectional and time-dependent associations.
• Existing methods may not adequately capture the complex dependencies inherent in such data.

Purpose of the Study:

• To develop a flexible statistical framework for joint modeling of multivariate longitudinal outcomes.
• To account for both the association between different outcomes at a specific time point and the association of repeated measurements over time for a single outcome.
• To address the issue of incomplete data in longitudinal studies.

Main Methods:

• A copula-based approach is employed for joint modeling of multivariate outcomes at each time point.
• A transition model is utilized to capture the association of longitudinal measurements over time.
• The missingness mechanism is assumed to be ignorable.
• Simulation studies are conducted using Gaussian, t, and Archimedean copulas.
• Akaike Information Criterion (AIC) is used for copula selection.

Main Results:

• The proposed method demonstrates flexibility in handling various marginal distributions.
• Simulation results show the performance of the approach under different scenarios.
• The Akaike Information Criterion effectively aids in selecting the most appropriate copula function.
• The approach is successfully applied to a real-world obesity dataset.

Conclusions:

• The copula-based joint modeling with a transition component provides a robust framework for multivariate longitudinal data.
• The method effectively handles complex associations and incomplete data.
• The approach offers a valuable tool for analyzing complex health-related datasets, such as obesity data.