I am using auc function from pROC package in R. Is it possible to use gtsummary package to put the results in table?

CodePudding user response：

You can turn any data frame into a table with gt::gt on which gtsummary is based, so, using the aSAH data set from pROC, you could do, for example:

library(pROC)
library(gt)
library(dplyr)

aSAH %>% 
    filter(gender == "Female") %>%
    roc(outcome, s100b) %>%
  {data.frame(
    predictor = as.character(.$call$predictor),
    "Area under curve" = auc(.), 
    "Confidence interval" =
        paste(as.character(round(ci.auc(b), 2))[c(1, 3)], collapse = " - "),
    check.names = FALSE)} %>%
  gt()

CodePudding user response：

There are a few ways to add the AUC. Below is an example adding it with the glance statistics from broom::glance()

library(gtsummary)
packageVersion("gtsummary")
#> [1] '1.6.0'

mod <- glm(response ~ age   trt, trial, family = binomial)

glance_with_auc <- function(x, ...) {
  df_glance <- broom::glance(x)
  
  # add AUC
  df_glance %>%
    dplyr::mutate(
      AUC = with(broom::augment(x), pROC::roc(response, .fitted))$auc %>% as.numeric()
    )
}
glance_with_auc(mod)
#> Setting levels: control = 0, case = 1
#> Setting direction: controls < cases
#> # A tibble: 1 × 9
#>   null.deviance df.null logLik   AIC   BIC deviance df.residual  nobs   AUC
#>           <dbl>   <int>  <dbl> <dbl> <dbl>    <dbl>       <int> <int> <dbl>
#> 1          229.     182  -113.  232.  241.     226.         180   183 0.576

tbl_regression(mod, exponentiate = TRUE) %>%
  add_glance_table(include = c(nobs, AIC, AUC), glance_fun = glance_with_auc) %>%
  as_kable() # convert to kable to display on SO
#> Setting levels: control = 0, case = 1
#> Setting direction: controls < cases

^{Created on 2022-05-15 by the reprex package (v2.0.1)}