biSurv: Non-parametric estimation of the bivariate survival function

In Jeepen/biSurv: Functions related to bivariate survival data

Non-parametric estimation of the bivariate survival function

Description

Non-parametric estimation of the bivariate survival function

Usage

biSurv(formula, data, gamma = NULL, maxIt = 100, method = "dabrowska")

Arguments

formula	a formula object, with the response on the left of a ~ operator, and the terms on the right. The response must be a survival object as returned by the Surv function. The RHS must contain a 'cluster' term
data	a data.frame containing the variables in the model
gamma	bandwidth of pruitt estimator. Can be ignored for other estimators.
maxIt	maximum number of iterations.
method	which estimator to use. 'dabrowska' (default), 'pruitt' or NPMLE

Details

many methods for bivariate survival data need an estimator for the bivariate survival function. This function makes it possible to choose between three estimators: 1. the Dabrowska estimator, which has the advantage of implying marginal survival functions given by the Kaplan-Meier, but gives negative probability mass to certain points, which is, of course, not wanted. 2. The Pruitt estimator which has positive probability mass everywhere and which has no problems with singly censored observations, but has a bandwidth, gamma (so strictly speaking not non-parametric). 3. The NPMLE estimator, which, as the name suggests, is a non-parametric MLE. It has the disadvantage that it doesn't converge for singly censored observations since probability mass has to be distributed over a line which almost surely has no observations lying on it.

Value

matrix with estimate of bivariate survival function

Author(s)

Jeppe E. H. Madsen <[e-mail]>

References

Hougaard, Philip. Analysis of multivariate survival data. Springer Science & Business Media, 2012.

van der Laan, Mark J. Modified EM-estimator of the bivariate survival function. Rijksuniversiteit Utrecht. Mathematisch Instituut, 1993.

Dabrowska, Dorota M. "Kaplan-Meier estimate on the plane." Annals of Statistics 16.4 (1988): 1475-1489.

See Also

print.biSurv plot.biSurv

Examples

library(survival)
data("diabetic")
biSurv(Surv(time,status)~cluster(id), data = diabetic)

Jeepen/biSurv documentation built on Sept. 30, 2023, 3:55 a.m.