rsurv() is a generic function for randomly drawing survival times for multiple
individuals very efficiently. The most flexible function is
rsurv.survival_curves(), which will randomly draw survival times from
survival curves.
rsurv(object, ...) # S3 method for survival_curves rsurv(object, n_rep = 1, ...)
| object | An object of the appropriate class. |
|---|---|
| ... | Further arguments passed to or from other methods. Currently unused. |
| n_rep | The number of times to randomly draw survival times per
individual ( |
A vector with randomly sampled survival times for individuals
(ids) in object repeated n_rep times. The returned vector is of length
length(unique(object$id)) times n_rep.
Survival times are randomly sampled from survival curves
by partitioning times into intervals and using the survival probabilities
to compute the probability of dying within each interval, conditional on
survival up until that interval. Death is simulated within each interval from
a Bernoulli distribution and a given individual's observed death time is
the time at the end of the first interval for which that individual is
simulated to have died. This process is repeated across all individuals.
The code is written in C++ so looping across individuals is fast.
A survival_curves object can be constructed using the function
survival_curves() or converted from an existing data frame using
as_survival_curves.data.frame().
# Survival curves for multiple subjects based on exponential distributions set.seed(21) N <- 1000 # 1000 individuals rates <- runif(N, .2, .5) sc <- example_survival_curves(n = N, rates = rates) # Random number generation of 1000 individuals from (1) arbitrary survival # curves and (2) exponential distributions sim_rsurv <- rsurv(sc) sim_rexp <- rexp(N, rate = rates) summary(sim_rsurv)#> Min. 1st Qu. Median Mean 3rd Qu. Max. #> 0.08333 0.83333 2.00000 2.82617 3.75000 26.33333#> Min. 1st Qu. Median Mean 3rd Qu. Max. #> 0.000361 0.824767 1.965348 3.032663 4.031558 25.952823# Random number generation of 1000 individuals replicated 2 times sim_rsurv <- rsurv(sc, n_rep = 2) summary(sim_rsurv)#> Min. 1st Qu. Median Mean 3rd Qu. Max. #> 0.08333 0.83333 2.00000 2.97721 4.08333 26.25000