Joint analysis of panel count data with an informative observation process and a dependent terminal event

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Survival Analysis

Background:

• Panel count data are common in clinical and observational studies.
• Informative observation processes and dependent terminal events (e.g., death) complicate data analysis.
• Existing models may not adequately address both informative observation and terminal events simultaneously.

Purpose of the Study:

• To propose a novel joint model for panel count data.
• To incorporate an informative observation process and a dependent terminal event.
• To provide robust statistical inference and model validation techniques.

Main Methods:

• Development of a joint model using two latent variables.
• Application of a class of estimating equations for parameter inference.
• Establishment of consistency and asymptotic normality for estimators.
• Introduction of a lack-of-fit test for model adequacy assessment.

Main Results:

• The proposed joint model effectively handles informative observation and dependent terminal events.
• Estimating equations yield consistent and asymptotically normal estimators.
• The lack-of-fit test is effective in assessing model adequacy.
• Simulation studies confirm the approach's utility in practical scenarios.

Conclusions:

• The novel joint model provides a powerful tool for analyzing complex panel count data.
• The statistical inference methods are sound and validated.
• The approach is applicable to real-world clinical trial data, as demonstrated by a bladder cancer example.