Joint modeling of generalized scale-change models for recurrent event and failure time data

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Data Modeling

Background:

• Recurrent event and failure time data are common in clinical studies.
• Existing models often rely on restrictive assumptions like the Poisson process or specific frailty distributions.

Purpose of the Study:

• To propose a flexible joint modeling approach for recurrent event and failure time data.
• To allow correlation between recurrent events and time-to-failure through a shared frailty term.
• To develop robust estimation methods without imposing strong distributional assumptions.

Main Methods:

• Joint modeling using generalized scale-change models for both processes.
• Incorporation of a shared frailty to account for correlation.
• Development of estimating equations for parameter estimation.
• Establishment of asymptotic properties for the estimators.

Main Results:

• The proposed joint model demonstrates flexibility by not requiring Poisson assumptions for recurrent events or parametric frailty distributions.
• Estimating equation approaches provide consistent parameter estimation.
• Simulation studies confirm the finite sample performance of the method.

Conclusions:

• The developed joint modeling framework offers a flexible and robust approach for analyzing correlated recurrent event and failure time data.
• The method is applicable to complex clinical data, as demonstrated by its use in a medical cost study of heart failure patients.