Temporal Supervised Contrastive Learning (Temporal-SCL) for Modeling Patient Risk Progression

Authors: Shahriar Noroozizadeh (snoroozi [at symbol] cs.cmu.edu), Jeremy C. Weiss, George H. Chen

This code accompanies the ML4H 2023 paper:

Shahriar Noroozizadeh, Jeremy C. Weiss, George H. Chen. "Temporal Supervised Contrastive Learning for Modeling Patient Risk Progression".
[arXiv]

Datasets

Three datasets are used in the experiments.

• Synthetic data: provided in this repo as data/data_synthetic.pickle; can be generated by running data/synthetic/data_generation.py.
• MIMIC III v1.4 data: publicly available at PhysioNet.
• ADNI data: can be downloaded from loni.

Experiments

• In notebooks/Tutorial.ipynb, an example of training Temporal-SCL and running inference on two different downstream tasks for the synthetic data can be found.
• The hyperparameters and arguments used for this experiment can be found in model\args_synthetic.json.

Citation

If you find the software useful, please consider citing the following paper:

@inproceedings{noroozizadeh2023temporal,
  title={Temporal Supervised Contrastive Learning for Modeling Patient Risk Progression},
  author={Noroozizadeh, Shahriar and Weiss, Jeremy C and Chen, George H},
  booktitle={Machine Learning for Health (ML4H)},
  pages={403--427},
  year={2023},
  organization={PMLR}
}