Estimation of the Conditional Survival Function for Ordered
Multivariate Failure Time Data

Method to implement some newly developed methods for the
estimation of the conditional survival function.

condSURV is an R package to implement some newly developed methods
for the estimation of the conditional survival function. The package implements
three nonparametric and semiparametric estimators for these quantities. The package also implements feasible estimation methods for these quantities conditionally on current or past covariate measures.

Other related estimators are also implemented in the package. One of these estimators is the Kaplan-Meier estimator typically assumed to estimate the survival function. A modification of the Kaplan-Meier
estimator, based on a preliminary estimation (presmoothing) of the
censoring probability for the survival time, given the available information is also implemented.

Installation

condSURV is available through both CRAN and GitHub.

This file documents software changes since the previous edition.

condSURV 1.0.0 (2016-10-28)

modifications were introduced in the package 'condSURV' to account for multiple prior events. The current version of the package allows the user to obtain the conditional survival function where the conditional event is multi-dimensional.

the survKMW, survLDM and survPLDM functions were merged in a single function survCOND (with a 'method' argument).

condSURV 2.0.0 (2016-12-05)

the survCOND and the survIPCW functions have been merged in just one (survCOND).

the survCOND function presents now a new argument formula similar to that in survfit function of the survival package. The cited formula object must be created, with the response on the left of a ′ ∼′ operator. The response must be a surv object which is obtained using the survCS function. Additionally, a single covariate (qualitative or quantita- tive) can be included in the right hand side of the formula allowing the estimation of survival probabilities conditionally on current or past covariate measures.

we have created a summary.surv function which can be used to to get the estimated values of the conditional probability at the desired values of “y” by means of the times argument.

the plot.surv function has been improved.

condSURV 2.0.1 (2016-12-20)

the 'surv' class has been removed. Now, the main class is 'survCS'.

the 'conf' argument of the survCOND function is TRUE by default.