#------preparation------ library("parfm") head(kidney) kidney$sex <- kidney$sex - 1 #----------------------- #------exponential-gamma------ mod <- parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist="exponential", frailty="gamma") mod ci.parfm(mod, level=0.05)["sex",] u <- predict(mod) plot(u, sort="i") #----------------------------- #------comparison of different models------ kidney.parfm <- select.parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist=c("exponential", "weibull", "gompertz", "loglogistic", "lognormal"), frailty=c("gamma", "ingau", "possta")) kidney.parfm plot(kidney.parfm) #------------------------------------------ #------exponential-inverse Gaussian------ parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist="exponential", frailty="ingau") #---------------------------------------- #------exponential-positive stable------ parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist="exponential", frailty="possta") parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist="exponential", frailty="possta", iniFpar=0.25) parfm(Surv(time, status) ~ sex + age, cluster="id", data=kidney, dist="exponential", frailty="possta", method="Nelder-Mead") #--------------------------------------- #------semi-parametric gamma------ coxph(Surv(time, status) ~ sex + age + frailty(id, distribution="gamma", eps=1e-11), outer.max=50, data=kidney) #---------------------------------