Nonparametric prediction of event times in randomized clinical trials.

摘要: In clinical trials with planned interim analysis, it can be valuable for logistical reasons to predict the times of landmark events such as 50th and 100th event. Bagiella Heitjan (Stat Med 2001; 20: 2055–63) proposed a parametric prediction model failure-time outcomes assuming exponential survival Poisson enrollment. When little is known about distributions interest, there concern that methods may biased inefficient if their underlying distributional assumptions are invalid. We propose nonparametric approaches make point interval predictions dates during course trial. obtain using Kaplan–Meier estimator extrapolate probability into future, selecting time when expected number equal number. To construct intervals, we use simulation strategy based on Bayesian bootstrap. Monte Carlo results demonstrate superiority no...