The Cox proportional hazards model has been used widely in genome-wide association (GWA) studies of censored time-varying phenotypes to investigate disease association expressed by the relative hazards. In this paper, we instead apply the so-called accelerated hazards model to explore a novel time scale-change genotypic association, which is not necessarily able to be identified by the traditional Cox models. Our application is motivated and demonstrated by a GWA study of hematopoietic stem cell transplantation cohort.
Keywords: Accelerated hazards; censoring; non-proportionality; semiparametric model.