SEMIPARAMETRIC REVERSED MEAN MODEL FOR RECURRENT EVENT PROCESS WITH INFORMATIVE TERMINAL EVENT

Area of Science:

• Biostatistics
• Survival Analysis
• Longitudinal Data Analysis

Background:

• Recurrent event processes are often influenced by terminal events, complicating analysis.
• Continuous observation is not always feasible, necessitating methods for discrete time points.
• Existing models struggle with informative terminal events in panel count data.

Purpose of the Study:

• To develop a robust statistical method for semiparametric regression of recurrent events with informative terminal events.
• To address challenges in discrete time observations and nuisance parameters.
• To accurately estimate baseline mean functions and covariate effects.

Main Methods:

• A semiparametric reversed mean model is proposed.
• A two-stage sieve likelihood-based estimation method is developed.
• The approach handles computational difficulties and is robust to Poisson process assumptions.

Main Results:

• The proposed two-stage estimator demonstrates consistency, convergence rate, and asymptotic normality.
• The method is validated through extensive simulation studies.
• The approach is successfully applied to real-world data from health and bladder tumor studies.

Conclusions:

• The novel method effectively analyzes recurrent events with informative terminal events in discrete time.
• The statistical properties of the estimator are theoretically established.
• The approach offers a valuable tool for analyzing complex longitudinal health data.