Two-stage pseudo maximum likelihood estimation of semiparametric copula-based regression models for semi-competing...

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Semi-competing risks data present unique challenges due to dependent censoring.
• Existing methods for analyzing such data can be computationally intensive or lack robustness.

Purpose of the Study:

• To propose a novel two-stage estimation procedure for copula-based models with semi-competing risks.
• To address dependent and independent censoring in survival data analysis.
• To develop a computationally efficient and robust statistical method.

Main Methods:

• Utilized a copula-based model with semiparametric transformation models for marginal survival functions.
• Implemented a two-stage estimation process: marginal terminal event estimation, followed by joint non-terminal and copula parameter estimation.
• Derived asymptotic properties and an analytic variance estimator for statistical inference.

Main Results:

• The proposed two-stage estimator demonstrated consistency and reduced computational cost compared to a one-stage approach.
• Simulation studies indicated superior finite-sample performance over existing two-stage methods.
• An R package (PMLE4SCR) was developed for practical implementation.

Conclusions:

• The proposed two-stage estimation procedure offers a robust and computationally efficient method for analyzing semi-competing risks data.
• This approach provides a valuable tool for researchers in biostatistics and related fields.
• The developed R package facilitates the application of this advanced statistical technique.