Semiparametric model averaging prediction for case K informatively interval-censored data

Abstract

Model averaging is an effective strategy for improving prediction accuracy that allows for parameter uncertainty. However, in survival analysis, most previous studies have focused on right-censored data. This paper addresses the issue of case K informatively interval-censored data, a common situation in practical applications such as medical follow-up studies. Despite its popularity, it has not received significant attention in the literature due to inherent challenges. To solve this problem, we construct a set of candidate joint models. We then propose a new semiparametric model averaging prediction (SMAP) method based on these joint candidate models. The weights in model averaging are determined by maximizing the pseudo-likelihood function. Under certain regular conditions, it is shown that the identified weights have asymptotically optimal properties. Simulation studies are conducted to assess the performance of the proposed method using different evaluation criteria. To illustrate, we apply the proposed method to the study of cardiac allograft vasculopathy.