Mean residual life cure models for right-censored data with and without length-biased sampling

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Survival data often includes a 'cured fraction' of individuals who will not experience the event of interest.
• Accurate modeling is crucial for understanding disease progression and treatment efficacy in clinical and epidemiological studies.
• Length-biased sampling, common in prevalent cohort studies, can complicate survival data analysis.

Purpose of the Study:

• To propose a novel semiparametric mixture cure model for right-censored survival data.
• To incorporate the proportional mean residual life model for uncured subjects and logistic regression for the cure rate.
• To develop robust estimation methods for data with and without length-biased sampling.

Main Methods:

• Developed estimating equations to estimate the semiparametric mean residual life mixture cure model.
• Proposed two distinct estimating equations for cure rate covariate effects, with a method for combining them to enhance efficiency.
• Established theoretical properties including consistency and asymptotic normality of the proposed estimators.

Main Results:

• The proposed model and estimation methods are effective for analyzing survival data with a cured fraction.
• Simulation studies confirmed the good finite sample performance of the developed estimators.
• The methods were successfully applied to real-world data from a melanoma clinical trial and a type 2 diabetes cohort study.

Conclusions:

• The developed semiparametric mixture cure model provides a flexible and effective framework for survival data analysis.
• The proposed estimation techniques offer reliable results, even in the presence of length-biased sampling.
• This approach enhances the ability to analyze covariate effects on both disease progression and cure probability in clinical and epidemiological research.