The C-index Multiverse

The C-index is a discrimination metric used in time-to-event prediction modeling. However, there are multiple implementations and choices that condition performance results. In this work, we study the effects of ties, censoring, time truncation and input transformations in the model ranking.

Given the multiple libraries used in this work, from R to python, we provide a Docker image with all code dependencies pre-installed.

Download files:

git clone https://github.com/BBolosSierra/CindexMultiverse.git
cd CindexMultiverse

Docker:

Install docker from https://www.docker.com

For Mac computers follow https://docs.docker.com/desktop/setup/install/mac-install/

docker pull ghcr.io/bbolossierra/cindex_multiverse_project:1.0.4

Interactively run Rstudio server:

docker container run --mount type=bind,source="$(pwd)",target=/home/rstudio/project -e PASSWORD=yourpassword -p 8787:8787 ghcr.io/bbolossierra/cindex_multiverse_project:1.0.4

Open your browser: http://localhost:8787

To log in; Username: rstudio Password: yourpassword (or whatever you set)

Use the Files pane to navigate to /CindexMultiverse

To remove the container:

docker rmi ghcr.io/bbolossierra/cindex_multiverse_project:1.0.4

R libraries

Package	Version
rhdf5	2.46.1
graph	1.80.0
Rgraphviz	2.46.0
parallelly	1.44.0
survival	3.8.3
data.table	1.17.2
stringr	1.5.1
dplyr	1.1.4
tidyr	1.3.1
gridExtra	2.3
ggalluvial	0.12.5
cowplot	1.1.3
patchwork	1.3.0
arrow	20.0.0
reticulate	1.42.0
rmarkdown	2.29
knitr	1.50
Hmisc	5.2.3
rms	6.8.1
prodlim	2025.4.28
pec	2023.4.12
riskRegression	2025.5.20
randomForestSRC	3.4.0
caret	7.0.1
doFuture	1.0.2
future	1.40.0
progressr	0.15.1
foreach	1.5.2
flexsurv	2.3.2
furrr	0.3.1
gtsummary	2.2.0
kableExtra	1.4.0
survivalmodels	0.1.191
tidyverse	2.0.0
cBioPortalData	2.14.2

Python packages

Package	Version
numpy	2.0.2
pandas	2.2.3
pycox	0.3.0
lifelines	0.30.0
scikit-survival	0.23.1
tqdm	4.67.1
numba	0.60.0