For prediction need, I want to save a coxph or cph model to be able use it in my function.
So my question is how to save the all coxph (or cph) object and then how to load it when needed?
The object will be used in the predictCox function (library riskRegression)
library(survival)
library(data.table)
library(riskRegression)
#### generate data ####
set.seed(10)
d <- sampleData(40,outcome="survival") ## training dataset
nd <- sampleData(4,outcome="survival") ## validation dataset
d$time <- round(d$time,1) ## create tied events
fit <- coxph(Surv(time,event)~X1 + X6,
data=d, ties="breslow", x = TRUE, y = TRUE)
## compute individual specific cumulative hazard and survival probabilities
fit.pred <- predictCox(fit, newdata=nd, times=3, se = TRUE, band = TRUE)
You can use the save function to store R objects in a file. If you want to store a single R object you should use the extension .rds and saveRDS function. On other hand, if you want to store several objects you should use the .Rdata file extension.
Examples
Single object:
saveRDS(model, file="model.rds")
model <- readRDS(file="model.rds")
In this case the object name is not stored, so you need to use the readRDS function and assign <- the object to a new variable in your environment.
Multiple objects:
save(model1, model2, df, file = "models_and_data.RData")
If you want to save all objects currently in your work environment you can call save.image instead:
save.image(file="project_image.RData")
In both cases all objects are saved with their original name, so you can load them into your environment with the load function without assigning new names:
load(file="models_and_data.RData")
Beware that some implementations of coxph make a copy of the data to model object (check if your model object has model in its structure, i.e model$model). If you are working with sensitive data and need to export models, make sure you are not exporting sensitive data hidden in the object.
Source: https://stat.ethz.ch/R-manual/R-devel/library/base/html/readRDS.html