It contains the code for predicting ASIA classification after a spinal cord injury.
The work herein has been published and the entirety of the research discussion can be found at: https://www.eneuro.org/content/10/1/ENEURO.0149-22.2022.
| Path | What does it contain? |
|---|---|
| ~/ | Everything (including this file) |
| ~/src/ | All code |
| ~/src/ml/ | All ML code |
| ~/src/ml/notebooks/ | ML ad hoc research code |
| ~/src/ml/modelling/ | ML training, testing, and feature importance code |
| ~/src/ml/modelling/pickles/ | ML model training outputs |
| ~/src/ml/modelling/plots/ | ML feature importance graphs |
| ~/src/ml/data/ | ML data handling and transformation code |
| ~/src/ml/data/csvs/ | Raw dataset |
| ~/src/ml/data/docs/ | Raw dataset documentation |
| ~/src/ml/data/utils/ | Commonly used utilities for data handling and transformation |
- Windows 10
- Git
- Anaconda or Miniconda (https://www.anaconda.com/)
- NSCISC Dataset (https://www.nscisc.uab.edu/Research/NSCISC_DatabasePublicUse)
- Clone repository:
git clone https://github.com/kapoor1992/spinal_cord_injury_recovery.git - Navigate to the shared directory:
cd src/ml - Create the conda environment:
conda env create --file environment_windows.yml - Activate the environment:
conda activate sci
- Copy
f1_public.csvfrom NSCISC intosrc/ml/data/csvs/ - Activate conda environment:
conda activate sci - Navigate to the modelling directory:
cd src/ml/modelling - Run model training (flags are optional):
python model_runs.py --interpret --drop-patients-who-worsened - If
--interpretwas used, verify thatimportance.pngcan be seen insrc/ml/modelling/plots/