A1446
Title: On confidence interval for correlation parameter in correlate gamma frailty models
Authors: Kana Takamatsu - Tokyo University of Science (Japan) [presenting]
Asanao Shimokawa - Tokyo University of Science (Japan)
Abstract: Survival analysis focuses on the time from a starting point, such as the initiation of surgery or treatment, until the occurrence of an event of interest. Many medical research datasets include information about subjects, such as age and sex, along with their survival times. One of the standard methods for analyzing the impact of these covariates on survival time is the Cox proportional hazards model. The genetic relationship between twins and families could potentially impact survival time. Therefore, a correlation frailty model with random effects is a method used to account for individual differences and non-independence within the same cluster. The distribution of random effects is often modeled using a gamma distribution. On the other hand, the properties of the estimated correlation parameter are not well-researched in many situations. The technique of constructing the confidence interval for the parameter is not well known. Therefore, the methods for estimating the confidence intervals of the correlation parameter in the correlated gamma frailty models are proposed using asymptotic distribution methods and bootstrap methods. These confidence intervals are compared through simulations and analysis of several examples.
