tahaceritli / miohmm

Code for Mixture of Input-Output Hidden Markov Models for Heterogeneous Disease Progression Modeling

Mixture of Input-Output Hidden Markov Models for Heterogeneous Disease Progression Modeling

This repository contains the code to reproduce the experiments in:

@InProceedings{ceritli2022,
    title = {Mixture of Input-Output Hidden Markov Models for Heterogeneous Disease Progression Modeling},
    author = {Ceritli, Taha and Creagh, Andrew P. and Clifton, David A.},
    booktitle={Proceedings of the 1st Workshop on Healthcare AI and COVID-19,
ICML 2022}

See https://arxiv.org/abs/2207.11846 for our latest arXiv submission.

Note that we extend the codebase in https://github.com/kseverso/DiseaseProgressionModeling-HMM and apply the data preprocessing steps in https://github.com/kseverso/Discovery-of-PD-States-using-ML. The repository is structured following the template provided in The Turing Way.

---

Contents: Introduction | Installation | Experiments | Contributing | License

---

Introduction

Installation

You can set up the necessary virtual environment via the following code:

conda create --name mIOHMM python=3.8.8
conda activate mIOHMM
pip install torch==1.10.0+cu113 torchvision==0.11.1+cu113 -f https://download.pytorch.org/whl/torch_stable.html
pip install scipy==1.6.1
pip install scikit-learn==0.24.1
pip install seaborn
pip install pandas
conda install -c anaconda ipykernel
python -m ipykernel install --user --name=mIOHMM

Experiments

You can reproduce the experiments via the command line as follows:

python -m experiments.synthetic
python -m experiments.real

You could also use the jupyter notebook to reproduce synthetic data experiments. We provide another jupyter notebook for real data experiments to reproduce the figures and tables used in the paper using the trained models.

Contributors

Taha Ceritli

Andrew P. Creagh

If you encounter an issue in mIOHMM, please open an issue or submit a pull request.

License

This work is licensed under the MIT license (code) and Creative Commons Attribution 4.0 International license (for documentation). You are free to share and adapt the material for any purpose, even commercially, as long as you provide attribution (give appropriate credit, provide a link to the license, and indicate if changes were made) in any reasonable manner, but not in any way that suggests the licensor endorses you or your use, and with no additional restrictions.

miohmm's Issues

Data for running the Real experiment

Hello,
There is no data in the data folder (either processed or raw) and I am not sure what structure should the data looks like for the preprocess function. Could you please at least upload the structure of Real data that your script need for running?
Thanks