CT-medication-errors

This repository provides the materials accompanying the paper
“Early Risk Stratification of Dosing Errors in Clinical Trials Using Machine Learning.”

The paper presents a dataset for the prediction of dosing errors in interventional clinical research, as well as a set of baseline models for this task.

The implementation of the models described in the paper is available in the aidose/baselines directory.

The Figures directory contains Jupyter notebooks used to generate the figures presented in the manuscript, as well as additional analyses (e.g., subgroup analyses).

The tests directory contains unit tests for the various components of the codebase.

Using the dataset

The dataset is hosted under the HuggingFace Hub at the following link:

https://huggingface.co/datasets/ds4dh/ct-dosing-errors

You can use it simply using the below commands:

from aidose.dataset import HF_HUB_REPO_ID

from datasets import load_dataset

ds = load_dataset(
    HF_HUB_REPO_ID,
    split="train"
)

print(ds)
print(ds.features)

Reproducing the dataset

In case you want to reproduce the dataset, you should first download an instance of the MedDRA terminology after having created an account under www.meddra.org and agreeing to their terms and conditions. This should then be placed under RESOURCES_DIR/MEDDRA/MedDRA_<MEDRA_VERSION>, where the default version is set to MedDRA_27_1_English.

After having cloned this repository, in a separate python environment, you should then install the aidose package with the command:

pip install -e .

This will install all the required dependencies.

You should then specify an environment variable DATASETS_ROOT, pointing to the root folder where you want to store the datasets, such as the CTGOV (>= 9 GBytes, which will be downloaded automatically by running the main script described next), as well as the final reproduced dataset CT-DOSING-ERRORS.

Once these requirements are satisfied, you may then proceed with the dataset creation script:

python3 aidose/dataset/main.py

The final dataset will be created under <DATASETS_ROOT>/<DATASET_NAME>/<DATASET_VERSION/.

Various global constants are set under constants.py, under the relevant packages within this repository.

License

This dataset is licensed under the CC BV 4.0 license.

Acknowledgements

This is part of the Innosuisse project "114.721 IP-ICT", titled "AIDosE: artificial intelligence methods to estimate and predict dosing errors in interventional clinical research". AIDosE is a joint collaboration between the University of Geneva, and Johnson & Johnson.