General Semiparametric Shared Frailty Model: Estimation and Simulation with frailtySurv

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Software

Background:

• Shared frailty models are essential for analyzing clustered survival data where observations are not independent.
• Existing statistical software may lack comprehensive options for various frailty distributions and robust estimation.

Purpose of the Study:

• Introduce the R package frailtySurv for simulating and fitting semi-parametric shared frailty models.
• Provide consistent estimators for multiple frailty distributions and their standard errors.

Main Methods:

• Implementation of semi-parametric consistent estimators for gamma, log-normal, inverse Gaussian, and power variance function frailty distributions.
• Utilizes asymptotic normality of parameter estimators for statistical inference (hypothesis testing, confidence intervals).

Main Results:

• Extensive simulations confirm the flexibility and correct implementation of the frailtySurv package.
• Case studies on Diabetic Retinopathy Study and hard drive failure data demonstrate practical applicability.

Conclusions:

• The frailtySurv package provides a valuable, flexible, and accurately implemented tool for analyzing clustered survival data.
• Facilitates robust statistical inference using various frailty distributions in real-world applications.