SEMIPARAMETRIC EFFICIENT ESTIMATION FOR SHARED-FRAILTY MODELS WITH DOUBLY-CENSORED CLUSTERED DATA.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Survival data often involves censoring, where the exact event time is unknown.
• Doubly-censored data, subject to both left and right censoring, presents unique analytical challenges.
• Frailty models are used to account for unobserved heterogeneity in survival data.

Purpose of the Study:

• To develop and evaluate a novel statistical approach for analyzing clustered, doubly-censored survival data.
• To extend the application of frailty models to scenarios involving both left and right censoring.
• To address computational challenges associated with existing methods for doubly-censored data.

Main Methods:

• A likelihood-based approach was employed to derive nonparametric maximum likelihood estimators (NPMLE).
• A new iterative algorithm was developed to compute the NPMLE for clustered, doubly-censored data.
• The algorithm's performance was assessed through simulations and applied to Hepatitis B family study data.

Main Results:

• The proposed algorithm effectively handles clustered survival data with both left and right censoring.
• It demonstrates improved performance over existing methods, particularly for the computationally challenging independent doubly-censored data case.
• Asymptotic properties and semi-parametric efficiency of the NPMLE were established, with Bootstrap consistency for standard errors discussed.

Conclusions:

• The new algorithm provides a robust and efficient solution for analyzing complex survival data structures.
• This work advances the application of frailty models in biostatistical research, particularly for clustered and doubly-censored outcomes.
• The methodology is validated by its successful application to real-world Hepatitis B epidemiological data.