Joint analysis of informatively interval-censored failure time and panel count data.

Area of Science:

• Biostatistics
• Survival Analysis
• Incomplete Data Analysis

Background:

• Interval-censored failure time and panel count data are common in medical and social sciences.
• Existing joint analysis methods often neglect the observation process, which can influence results.
• The interdependency between data types and observation process requires a comprehensive modeling approach.

Purpose of the Study:

• To develop a three-component joint model for interval-censored failure time, panel counts, and the observation process.
• To address the limitations of previous analyses by incorporating the observation process.
• To provide a unified statistical framework for analyzing these complex data types.

Main Methods:

• Utilized a three-component joint model incorporating gamma and distribution-free frailties.
• Employed a sieve maximum likelihood estimation approach.
• Applied Bernstein polynomial approximation for parameter and baseline hazard function estimation.

Main Results:

• Established asymptotic properties for the proposed estimators.
• Simulation studies confirmed the procedure's effectiveness in practical scenarios.
• Successfully applied the method to a real-life cardiac allograft vasculopathy dataset.

Conclusions:

• The proposed joint model effectively analyzes interval-censored failure time and panel count data, considering the observation process.
• The sieve maximum likelihood and Bernstein polynomial approximation provide reliable estimation.
• The method demonstrates practical utility in medical research, as shown by the cardiac allograft vasculopathy study application.