Frailty Model Approach for the Clustered Interval-Censored Data with Informative Censoring

• Activity Number: 80
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Jinheum Kim*
• Companies: University of Suwon
• Keywords: Non-informative censoring ; Frailty model ; EM algorithm ; Gauss-Hermite approximation ; Interval-censored data ; Diabetic retinopathy data

Abstract

Interval-censored outcomes are often arisen when the time to an event of interest is known to have occurred within a specific time period. In many clinical trials in conjunction with periodic follow-up, each subject or individual is supposed be observed through several examinations. Most of existing methodologies with regression analysis are extended and developed under the assumption of non-informative censoring mechanism. However, there will be situations that this assumption does not hold. Subsequently, it is not possible to test the dependence or independence assumption of the censoring mechanism. One remedy to circumvent these difficulties is to impose extra assumptions or additional modelling. In this article, we employ the Cox proportional hazards models with a frailty effect being incorporated with interval-censored data for which there is the dependence between the failure time and visiting times. Parameter estimation was achieved using EM algorithm and also simulations were performed to investigate its finite-sample properties. *This research was supported by Basic Science Research Program through the NRF of Korea funded by the MSIP (No. NRF-2014R1A1A2056869)