Project Details
Projekt
Tackling Challenges in Time-to-Event Analyses via Boosting Distributional Copula Regression
Subject Area
Epidemiology and Medical Biometry/Statistics
Term
since 2019
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 428239776
Survival and other time-to-event outcomes are among the most prominent endpoints in clinical trials. Biostatistical methods for analyzing such data differ significantly from conventional techniques due to the presence of censoring, which occurs when the exact time of an event is unknown. Classical methods for time-to-event data often incur assumptions that make it difficult to accurately model the data generating process and can lead to incorrect conclusions drawn from the data. Beyond addressing censoring, it is also important to identify relevant or informative factors for the respective event time endpoints and to understand their dependence structure. This project is concerned with developing sophisticated regression and variable selection techniques tailored to the joint analysis of complex data structures where the outcome of interest is comprised of multiple, potentially inter-dependent time-to-event or survival endpoints. We devise advanced copula-based approaches for high-dimensional datasets in order to tackle biomedical research questions related to time-to-event endpoints. The proposed methods shall yield interpretable models of potentially complex association and dependence structures as well as high predictive performance. All of the developed methods will be implemented in freely available software in order to broaden the applicability as well as to foster transparent and reproducible research.
DFG Programme
Research Grants
International Connection
Belgium, United Kingdom
Cooperation Partners
Professor Dr. Miguel de Carvalho; Professorin Dr. Ingrid Van Keilegom
