Simulate survival times from standard parametric survival
distributions (exponential, Weibull, Gompertz), 2-component mixture
distributions, or a user-defined hazard, log hazard, cumulative hazard,
or log cumulative hazard function. Baseline covariates can be included
under a proportional hazards assumption.
Time dependent effects (i.e. non-proportional hazards) can be included by
interacting covariates with linear time or a user-defined function of time.
Clustered event times are also accommodated.
The 2-component mixture distributions can allow for a variety of flexible
baseline hazard functions reflecting those seen in practice.
If the user wishes to provide a user-defined
hazard or log hazard function then this is possible, and the resulting
cumulative hazard function does not need to have a closed-form solution.
Note that this package is modelled on the 'survsim' package available in
the 'Stata' software (see Crowther and Lambert (2012)
< http://www.stata-journal.com/sjpdf.html?articlenum=st0275> or
Crowther and Lambert (2013) ).

News

simsurv 0.2.3-9000 (X/X/2019)

Current development version

simsurv 0.2.3 (1/2/2019)

New features

The rootfun argument has been added. This allows the user to apply any transformation to each side of the root finding equation. The default is rootfun = log which corresponds to uniroot solving -H(t) - log(u) = 0.

The rootsolver argument has been added. This allows the user to choose between stats::uniroot or BB::dfsane for the root finding.

simsurv 0.2.2 (18/5/2018)

New features

The uniroot function (called internally by simsurv) now solves - H(t) - log(u) = 0 instead of exp(-H(t)) - u = 0, i.e. it is on log scale.

simsurv 0.2.1 (16/5/2018)

Bug fixes

The interval argument of simsurv supports a lower limit of zero (thanks to Alessandro Gasparini, @ellessenne).

simsurv 0.2.0 (6/3/2018)

New features

Added technical and example usage vignettes

Added user-specified cumulative hazard or log cumulative hazard

Added analytical forms for the inverted survival function when generating survival times from standard distributions (instead of using numerical root finding). This has lead to about a 5-fold increase in speed when simulating event times from standard parametric distributions.