This is the repository for the His pacing machine learning project.

The open access manuscript is available here

train.py - Train the 3 CNNs via 3-fold cross-validation across the training set

test.py - Evaluate the ensembled 3 CNNs on the testing dataset

vis.py - Create the saliency maps

The pre-trained networks are available in output/models/

Saliency maps for the entire testing set are available in output/vis/

The Python version used is 3.8 and the following packages are required (all installed by with conda via the conda-forge channel)

• matplotlib
• pytorch v1.5 & torchvision (pytorch conda channel)
• pyyaml
• numpy
• pandas
• scikit-image
• scikit-learn
• scipy
• statsmodels
• tqdm
• xlrd