New Computing Method for Joint Modeling Longitudinal and Survival Data

• Activity Number: 134
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Xiaoyu Liu*+ and Runze Li
• Companies: Penn State and Penn State
• Keywords: joint modeling ; longitudinal data ; survival data ; EM algorithm ; design of experiment

Abstract

Joint models (JM) of longitudinal and survival data have been developed in the past two decades as tools to characterize the two types of data simultaneously via the shared information. In this paper we study the computational challenges which have always been the implementation obstacles for complex JM settings. We discuss the three most popular computing techniques in JM context. Since these approaches are inadequate to fit the JM with large dimensions of random effects, we introduce a new computing method, design of experiments-based interpolation techniques (DoIt, Joseph (2012)), into the JM framework. DoIt is faster in speed, more stable in computation and thus more capable of solving the problem. Simulations and real data analysis are conducted to demonstrate the good performance and practical usefulness of this new method.