A Bayesian semiparametric multivariate joint model for multiple longitudinal outcomes and a time-to-event

Area of Science:

• Biostatistics
• Medical Statistics
• Survival Analysis

Background:

• Renal graft failure is a critical outcome in kidney transplantation.
• Accurate prediction of graft survival requires models that integrate longitudinal health markers with time-to-event data.
• Existing joint models may lack the flexibility to capture complex patient trajectories.

Purpose of the Study:

• To propose a novel semiparametric multivariate joint model for analyzing renal graft failure.
• To enhance model flexibility by employing nonparametric approaches for key components.
• To provide practical guidance on parameterization strategies for relating longitudinal data to survival outcomes.

Main Methods:

• Utilized a spline-based approach for subject-specific longitudinal evolutions.
• Modeled the baseline risk function using a piecewise constant approach.
• Employed a Dirichlet Process prior formulation for the distribution of latent terms.
• Investigated three main families of parameterizations for linking longitudinal processes to survival outcomes.

Main Results:

• The proposed semiparametric joint model offers increased flexibility in capturing complex data structures.
• Nonparametric components allow for data-driven modeling of longitudinal trends and survival risks.
• The study provides a framework for selecting appropriate parameterizations based on practical considerations.

Conclusions:

• The developed joint model provides a robust and flexible framework for analyzing renal graft failure.
• The nonparametric approach enhances the ability to model subject-specific trajectories and survival.
• Guidance on parameterization aids clinicians in applying these models effectively for improved patient management.