A0629
Title: Classification of competing risks under a semiparametric density ratio model with transition of markers
Authors: Huijun Jiang - University of North Carolina at Chapel Hill (United States) [presenting]
Quefeng Li - University of North Carolina - Chapel Hill (United States)
Jessica Lin - University of North Carolina at Chapel Hill (United States)
Feng-Chang Lin - University of North Carolina - Chapel Hill (United States)
Abstract: The cause of failure for competing risk data may not always be observable, which imposes additional challenges for estimating the risk of the primary event of interest. In many infectious diseases, episodes of recurrence may arise through a relapse of the initial infection or a new infection. Identifying the true cause of infection is essential to aid in choosing the appropriate treatment. Using time-to-event information, a novel method is presented for classifying the latent cause of failure under a semiparametric density ratio model. The expectation-maximization (EM) algorithm estimates unknown parameters, including the marginal probabilities of the patient-specific causes of failure. In addition, transition likelihoods between covariates at the baseline and at the time of event occurrence are used to provide a better, at least not worse, classification result. The simulation experiments are performed under various scenarios, such as sample size, censoring rate, and approximation methods for estimating the baseline hazard function. The numerical results show that the proposed classifier performs well under all settings. The proposed method is also applied to Cambodia's P. vivax malaria data, classifying recurrent malaria infections as relapse or reinfection.
