Fisher information for two gamma frailty bivariate Weibull models

Area of Science:

• Statistics
• Biostatistics
• Survival Analysis

Background:

• Asymptotic properties of frailty models for multivariate survival data are not well understood.
• Frailty models are crucial for analyzing correlated survival times in various fields.
• Understanding asymptotic behavior is essential for reliable statistical inference.

Purpose of the Study:

• To investigate the asymptotic properties of bivariate gamma frailty models.
• To derive the Fisher information in two specific bivariate gamma frailty models.
• To compare the Fisher information under different distributional assumptions.

Main Methods:

• Derivation of Fisher information for the standard bivariate gamma frailty model with Weibull conditional survival.
• Derivation of Fisher information for a bivariate gamma frailty model with Weibull marginal distributions.
• Comparative analysis of the derived Fisher information quantities.

Main Results:

• The Fisher information was successfully derived for both considered bivariate gamma frailty models.
• The derivation provides a basis for understanding the asymptotic behavior of these models.
• The comparison highlights differences in information content based on model specification.

Conclusions:

• The study provides key insights into the asymptotic properties of bivariate frailty models.
• The derived Fisher information is valuable for statistical inference and model selection.
• Further research can build upon these findings to develop more robust survival analysis methods.