Regression models using parametric pseudo-observations.

摘要: Pseudo-observations based on the nonparametric Kaplan-Meier estimator of survival function have been proposed as an alternative to widely used Cox model for analyzing censored time-to-event data. Using a spline-based has some potential benefits over approach in terms less variability. We propose define pseudo-observations flexible parametric and use these analysis regression models estimate parameters related cumulative risk. report results simulation study that compares empirical standard errors estimates various settings. Our simulations show situations there is substantial gain reduced variability using compared with pseudo-observations. The can be measured reduction error by up about one third; corresponding additional 125% larger sample size. illustrate method brief data example.