ezcox: Easily Process a Batch of Cox Models
In ezcox: Easily Process a Batch of Cox Models

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

The goal of ezcox is to operate a batch of univariate or multivariate Cox models and return tidy result.

Installation

You can install the released version of ezcox from CRAN with:

install.packages("ezcox")

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("ShixiangWang/ezcox")

Visualization feature of ezcox needs the recent version of forestmodel, please run the following commands:

remotes::install_github("ShixiangWang/forestmodel")

Example

This is a basic example which shows you how to get result from a batch of cox models.

library(survival)
library(ezcox)
data(lung)
head(lung)

# Build unvariable models
ezcox(lung, covariates = c("age", "sex", "ph.ecog"))

# Build multi-variable models
# Control variable 'age'
ezcox(lung, covariates = c("sex", "ph.ecog"), controls = "age")

Run parallelly

For parallel computation, users can use ezcox_parallel(). This function has same arguments as ezcox(). For variables < 200, this function is not recommended.

ezcox_parallel(lung, covariates = c("sex", "ph.ecog"), controls = "age")

Filter

Sometimes, we may need to filter result from multi-variable models.

lung$ph.ecog = factor(lung$ph.ecog)
zz = ezcox(lung, covariates = "sex", controls = "ph.ecog")

zz

# At default, it will drop all control variables
filter_ezcox(zz)

# You can specify levels to filter out
filter_ezcox(zz, c("0", "2"))
filter_ezcox(zz, c("0", "2"), type = "contrast")
filter_ezcox(zz, c("0", "2"), type = "ref")

# More see ?filter_ezcox

Get models

Get raw models may help users understand the detail and do further visualization.

zz = ezcox(lung, covariates = c("sex", "ph.ecog"), controls = "age", return_models=TRUE)
mds = get_models(zz)
str(mds, max.level = 1)

Show models

show_models(mds)
# Set model names
show_models(mds, model_names = paste0("Model ", 1:2))
# Merge all models and drop control variables
show_models(mds, merge_models = TRUE, drop_controls = TRUE)