Nonparametric maximum likelihood estimation of the survival function using current lifetime data
Nonparametric maximum likelihood estimation of the survival function using current lifetime data
Where did the research take place?
The study site has not been established. Author addresses may differ from where the research occurred.
A plain-language reading has not been prepared for this paper yet.
Original abstract
Abstract An issue when estimating the failure time survival function is how to set up a prevalent cohort study infrastructure to follow subjects after enrollment. This problem can be circumvented through the well‐known Grenander density estimator using current lifetime observations only. However, Grenander's estimator suffers from inconsistency at the origin, making it unsuitable for survival function estimation. We present a direct approach to this estimation problem, which circumvents the inconsistency at the origin without the necessity of penalization, by proposing a nonparametric maximum likelihood survival function estimator defined over a set of evenly spaced points. Through simulation studies, we show that the estimator performs well relative to the nonparametric maximum likelihood estimator defined over the observed current lifetimes. We also apply the proposed estimator to model survival with Parkinson's disease using data from the Canadian Open Parkinson Network.