IDWSDS 2024 - S16: New developments in dependent censoring with unknown association

In survival analysis it is commonly assumed that the survival time T and censoring time C are stochastically independent. Most commonly used models and methods (like Kaplan-Meier, Cox model, AFT model, log-rank tests,...) are making use of this assumption. However, there are situations in practice where this assumption might be violated. Consider for instance the situation where some patients leave a medical study for reasons related to their health, which will then indirectly be related to their survival time. Recent research starting with Czado and Van Keilegom (2023, Biometrika) has shown that by making use of copulas to describe the relation between T and C, their association can be identified under certain conditions, which is an important step forward. In this session three talks will be given that deal with dependent censoring and that are building further on the aforementioned paper.