Dependent censoring regression models for survival multivariate data. These models are based on extensions of the frailty models, capable to accommodating the dependence between failure and censoring times, with Weibull and piecewise exponential marginal distributions.
You can install the development version of DepCens from GitHub with:
#install.packages("devtools")
#devtools::install_github("GabrielGrandemagne/DepCens")This is a basic example which shows you how to solve a common problem:
library(DepCens)
#KidneyMimic is our simulated data frame
delta_t <- ifelse(KidneyMimic$cens==1,1,0)
delta_c <- ifelse(KidneyMimic$cens==2,1,0)
fit <- dependent.censoring(formula = time ~ x1 + x2 | x3 + x1, data=KidneyMimic, delta_t=delta_t,
delta_c=delta_c, ident=KidneyMimic$ident, dist = "weibull")
summary_dc(fit)
#>
#> Weibull approach
#>
#> Name Estimate Std. Error CI INF CI SUP p-value
#> Alpha 1.388009 0.3922203 0.6192571 2.156761 0.0004019
#> Sigma 0.6447018 0.2315937 0.1907782 1.098625
#>
#> Coefficients T:
#>
#> Name Estimate Std. Error CI INF CI SUP p-value
#> x1 0.08098522 0.02080334 0.04021067 0.1217598 9.905e-05
#> x2 -1.406800 0.2398661 -1.876938 -0.9366627 4.493e-09
#>
#> Coefficients C:
#>
#> Name Estimate Std. Error CI INF CI SUP p-value
#> x3 0.230895 0.1888372 -0.1392259 0.601016 0.2214
#> x1 0.1945837 0.03807359 0.1199595 0.2692079 3.209e-07
#>
#> ----------------------------------------------------------------------------------
#>
#> Information criteria:
#>
#> AIC BIC HQ
#> 405.4663 435.1512 417.4793You can also plot the survival function
plot_dc(fit, scenario = "t")