Copula-based analysis of dependent current status data with semiparametric linear transformation model

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Dependent censoring in current status data poses challenges for standard regression models.
• Existing copula-based methods often struggle with model and parameter identification.
• Accurate analysis is crucial in fields like epidemiology and tumorigenicity experiments.

Purpose of the Study:

• To propose a novel copula-based regression analysis for current status data with dependent censoring.
• To address limitations in model and parameter identification of existing methods.
• To develop a flexible semiparametric approach where the association parameter is unspecified.

Main Methods:

• Utilizing a general class of semiparametric linear transformation models.
• Employing parametric copulas to model the dependence structure.
• Implementing sieve maximum likelihood estimation with Bernstein polynomial approximation for nonparametric functions.

Main Results:

• Demonstrated identifiability of the proposed semiparametric model under regularity conditions.
• Established asymptotic consistency and normality of the developed estimators.
• Validated the method's effectiveness through extensive simulations and a real data application.

Conclusions:

• The proposed copula-based method offers a robust solution for analyzing current status data with dependent censoring.
• The approach improves upon existing methods by addressing identifiability issues.
• The technique is practically applicable and effective in real-world scenarios.