ICAN (https://doi.org/10.1093/neuros/nyw135) is a French research program aimed at better understanding the pathophysiology of intracranial aneurysm (IA). One of the addressed challenges is to develop diagnostic and predictive tools addressing IA rupture risk. For computational reproducibility, we provide here a simulated clinical dataset tooled with Python and R notebooks. This material supports the paper "Location of intracranial aneurysms is the main factor associated with rupture in the ICAN population" published at https://jnnp.bmj.com/content/92/2/122.abstract.
Contacts
- Olivia Rousseau, olivia.rousseau@univ-nantes.fr
- Matilde Karakachoff, matilde.karakachoff@chu-nantes.fr
- Alban Gaignard, alban.gaignard@univ-nantes.fr
We are gratefull to the MyBinder service. It allows to launch and configure virtual machines with the required software environment. Dependencies are specified in the environment.yml file. You can then interact with data analyis pipelines made available through jupyter notebooks .
- Clinical-data-simulator.ipynb: shows how we produce simulated data. Numpy random functions are extensively used to mimic virtual subjects with probablity distributions close to whats is observed in the real ICAN data collection. Explored varaibles are detailed in sim-data.md .
- Table-Baseline-Characteristics.ipynb: shows how variables are represented in the ruptured and unruptured sub-populations and hwo p-values are computed. .
- Factor-Analysis-for-Mixed-Data.ipynb: shows how the FAMD dimensionality reduction method is applied to our dataset.
- Logistic-Regression-Model.ipynb
- Random-Forest-Model.ipynb
- Models-comparison.ipynb
We also provider a Docker container hosting the packaged software environment and the jupyter notebooks.
As soon as docker is installed, launch the command docker pull albangaignard/ican-ml
to retrieve the container, and docker run -p 8888:8888 -i -t albangaignard/ican-ml
to run it. This will launch a jupyter notebook on your local computer on port 8888. Finally just browse the url specified in the terminal to enter into the notebook environment.