/surgVAE

Primary LanguagePythonMIT LicenseMIT

surgicalVAE

The implementation of surgicalVAE (surgical Variational Autoencoder),

the innovative and highly accurate VAE-based Perioperative Prediction Framework (Semi-supervised and Multi-tasking model),

capable of simultaneous prediction across N = 6 different crucial outcomes in high-risk cardiac surgery.

image

The repo contains training and testing codes.

Trained model files can be downloaded from here: https://figshare.com/s/e44f0120502b01583ba2. Trained under 5-fold cross-validation settings specified in the paper.