Modeling clustered long-term survivors using marginal mixture cure model.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Modeling right-censored clustered survival data with cured fractions is complex.
• Marginal mixture cure models are of recent interest for such data.
• Existing generalized estimating equation approaches require enhancement.

Purpose of the Study:

• To propose a semiparametric marginal mixture cure model for right-censored clustered survival data.
• To extend existing generalized estimating equation methods with an unbiased estimating equation for latency regression parameters.
• To establish large sample properties and evaluate finite sample performance of the proposed estimators.

Main Methods:

• Development of a semiparametric marginal mixture cure model.
• Extension of generalized estimating equations with a novel unbiased estimating equation.
• Theoretical establishment of large sample properties for regression effect estimators.
• Simulation studies to assess finite sample performance.

Main Results:

• The proposed method provides unbiased estimating equations for regression parameters in both incidence and latency.
• Large sample properties of regression effect estimators are theoretically established.
• Simulation studies demonstrate the finite sample performance of the proposed estimators.

Conclusions:

• The developed semiparametric marginal mixture cure model offers a robust approach for analyzing complex survival data.
• The extended generalized estimating equation method provides reliable estimators for incidence and latency.
• The method is validated through simulation studies and illustrated with real-world medical data.