M.A. Nicolaie focuses in this thesis on inference in survival models for survival data with competing risks. The author introduces a new approach to competing risks data, called vertical modeling.... Show moreM.A. Nicolaie focuses in this thesis on inference in survival models for survival data with competing risks. The author introduces a new approach to competing risks data, called vertical modeling. It is built on natural observable quantities in competing risks, that is, it quantifies 1. the chance that a failure occurs, irrespective of its cause and 2. conditionally that a failure occurred, it quantifies the risk that the event of failure is ascertained to a certain type of failure. Another appealing feature of vertical modeling which is discussed is that it deals with competing risks when missing causes of failure occur. Next, the author tackles the topic of dynamic prediction in competing risks, a topical subject nowadays. She uses two different approaches, one which is based on modeling the cause-specific hazards and one which is based on modeling the dynamic pseudo-observations associated to the cumulative incidence functions. The results presented in this thesis provide key messages on the use of competing risks methods in different fields such as epidemiology, medicine, demography.M.A. Nicolaie focuses in this thesis on inference in survival models for survival data with competing risks. The author introduces a new approach to competing risks data, called vertical modeling. It is built on natural observable quantities in competing risks, that is, it quantifies 1. the chance that a failure occurs, irrespective of its cause and 2. conditionally that a failure occurred, it quantifies the risk that the event of failure is ascertained to a certain type of failure. Another appealing feature of vertical modeling which is discussed is that it deals with competing risks when missing causes of failure occur. Next, the author tackles the topic of dynamic prediction in competing risks, a topical subject nowadays. She uses two different approaches, one which is based on modeling the cause-specific hazards and one which is based on modeling the dynamic pseudo-observations associated to the cumulative incidence functions. The results presented in this thesis provide key messages on the use of competing risks methods in different fields such as epidemiology, medicine, demography. Show less
Nicolaie, M.A.; Houwelingen, J.C. van; Witte, T.M. de; Putter, H. 2013