- Install python3.7
- Install most up-to-date version of wfdb (2.2.1). Ideally from github as the deployed version has some issues with reading and writing annotation/sample files.
- Install all packages in requirements.txt
- The host should have docker installed. Was programmed with docker-rootless.
- Download if necessary datasets from physionet under /mnt/dsets/physionet (aami-ec13 afdb edb mitdb nsrdb nstdb svdb vfdb)
- start app with the following commands
export FLASK_ENV=development
export FLASK_DEBUG=0
echo $FLASK_APP
echo $FLASK_ENV
echo $FLASK_DEBUG
python3.7 -u -m flask run --host=0.0.0.0 -p 5001
You can find examples under test/algorithms. Each folder in this directory can be zipped and uploaded. Every .zip file needs to contains a shell script called "setup.sh" which will then be executed. In this script you need to deploy and execute your algorithm on the data saved under /data/. Please save your predictions in the format of physionet https://physionet.org/about/database/) under the /pred/ path.