A1583
Title: Survival tree for current status data with dependent censoring
Authors: Yang-Jin Kim - Sookmyung Women University (Korea, South) [presenting]
Abstract: In survival analysis, the independent censoring assumption is commonly used for inference. However, this assumption is often violated in observational studies, and inference ignoring dependency produces invalid and biased results. A survival tree method for current status data with dependent censoring is proposed by extending several survival forests such as cforest and randomForestSRC. Current status data arise in cross-sectional studies where each subject is examined to determine whether an event of interest occurs at a given censoring time. The copula graphic estimator is applied to construct the evaluation and splitting criterion. Several methods are compared to evaluate predictive performance.
