mhmcc / ed-interaction-mapping

Mapping patient interactions in psychiatric presentations to a tertiary emergency department using network science and machine learning

Mapping patient interactions in psychiatric presentations to a tertiary emergency department

Jupyter notebook and Python code for the analysis of patient trajectories in a hospital emergency department for individuals presenting with suspected risk of suicide or self-harm, as published in our study:

McCullough MH, Small M, Jayawardena B, Hood S (2024) Mapping clinical interactions in an Australian tertiary hospital emergency department for patients presenting with risk of suicide or self-harm: Network modeling from observational data. PLOS Medicine 21(1): e1004241. https://doi.org/10.1371/journal.pmed.1004241

 

Overview

This repository contains:

• The Jupyter notebook ED-interaction-mapping.ipynb which runs the complete pipeline for analysing patient trajectory/interaction data, generating results and rendering figures for our published study.
• YAML file ED-interaction-mapping.yml for installing the conda environment.
• CSV files for the study data and lookup tables for system agents (doctors, patients etc.) and referral decisions.

Installation and usage

To run the notebook:

1. Clone the repository.
2. Install the environment with conda using the included YAML file.

conda env create -f ED-interaction-mapping.yml

1. Run all cells in order.

Data availability

The data for this project are stored within this repository in data.csv. Note that date information has been altered for data privacy.

References

Please see the companion paper linked above for details of the methods and packages used in this code.

Citing this code

If you use or adapt our code or methods in your research, please cite the companion paper linked above or as shown here in BibTeX format:

@article{10.1371/journal.pmed.1004241,
    doi = {10.1371/journal.pmed.1004241},
    author = {McCullough, Michael H. AND Small, Michael AND Jayawardena, Binu AND Hood, Sean},
    journal = {PLOS Medicine},
    publisher = {Public Library of Science},
    title = {Mapping clinical interactions in an Australian tertiary hospital emergency department for patients presenting with risk of suicide or self-harm: Network modeling from observational data},
    year = {2024},
    month = {01},
    volume = {21},
    url = {https://doi.org/10.1371/journal.pmed.1004241},
    pages = {1-21},
    number = {1},
}

This repository was archived on Zenodo upon publication:

Questions

Please email [email protected] if you have questions about the code.