Skip to content

Latest commit

 

History

History
53 lines (39 loc) · 1.95 KB

readme.md

File metadata and controls

53 lines (39 loc) · 1.95 KB
Raw

flexsurvcure

Travis build status codecov

The development repository for flexsurvcure, an R package for parametric mixture and non-mixture cure models. Flexsurvcure is based on flexsurv, the R package for parametric survival modelling.

Installation (development version)

install.packages("flexsurv")
install.packages("devtools")
devtools::install_github('jrdnmdhl/flexsurvcure')

Supported Distributions

All of the built-in distribution with flexsurvreg are supported, though some currently have issues with convergence and numerical instability. The following distributions currently seem reliable:

  • Exponential (exp)
  • Weibull (weibull, weibullPH)
  • Lognormal (lnorm)
  • Log-Logistic (llogis)

The following distributions are supported, but may not be reliable:

  • Gompertz (gompertz)
  • Generalized Gamma (gengamma, gengamma.orig)
  • Generalized F (genf, genf.orig)

Custom distributions can also be used by passing a distribution list (see flexsurv examples).

Fitting a mixture cure model

mixture = flexsurvcure(Surv(rectime,censrec)~group, data=bc, dist="weibullPH", link="logistic", mixture = T)
plot(mixture)

Fitting a non-mixture cure model

non_mixture = flexsurvcure(Surv(rectime,censrec)~group, data=bc, dist="weibullPH", link="loglog", mixture = F)
plot(non_mixture)

Covariates on parameters other than cure fraction

non_mixture_covarite_scale = flexsurvcure(Surv(rectime,censrec)~group, data=bc, anc=list(scale=~group), dist="weibullPH", link="loglog", mixture = F)
plot(non_mixture_covarite_scale)