B1486
Title: On survival trees with competing risks
Authors: Asanao Shimokawa - Tokyo University of Science (Japan) [presenting]
Etsuo Miyaoka - Tokyo University of Science (Japan)
Abstract: Although there is extensive research on survival trees, almost all of them assume only one event risk. However, we should consider the mutual effects of various risks (competing risks) if there are several event risks. Therefore, the goal is to construct a tree-structured model that considers the effects of competing risks on the survival time of interesting risks. To achieve this goal, we propose a criterion based on the finite mixture models of the exponential distributions. We evaluate the performance of this criterion through simulation studies. In addition to this, we show the results of the survival tree obtained from actual data.
