Tactical tool for CUH Covid configuration planning

An interactive dashboard used by decision makers at a university hospital to evaluate strategies for ward re-configuration* during the 2020 coronavirus pandemic.

This tool helped hospital management develop their understanding of the impact their decisions and their assumptions would have on the hospital's ability to treat a varying "wave" of Covid patients along with their capacity to continue to serve their "normal" patient demand.

*Note: ward re-configuration refers to whether a ward is allocated to patients who: (1) definitely have Covid; (2) potentially have Covid; and (3) definitely do not have Covid. These classifications were termed Red, Amber, and Green, respectively.

Full screenshot of the interactive tool

Tool architecture

The tool is comprised of four stages:

1. Admissions numbers generation
2. Conversion of those admissions numbers to bed occupancy
3. Trigger signal evaluation
4. Ward configuration output

Further explanation is contained within the full report, available on request. This work was undertaken as part of my MPhil dissertation at the University of Cambridge.

---

Development

0. Getting setup

This project uses an Anaconda environment that contains all the required Python package dependencies.

0.1. Install Anaconda

First, you'll need to install Anaconda. Follow the instructions their website.

0.2. Create a conda environment for the project

To get started, just run the included setup.sh bash script (note permissions issues may require you to run the commands manually):

# create the conda environment and install dependencies
conda create --name cuhvid --file requirements.txt
# add local modules to environment
conda develop .

This creates a conda environment called cuhvid and installs all the dependencies. It then adds the project folder to the conda environment to allow the local package (also called cuhvid) to be used.

1. Serve the interactive dashboard on local machine

You can serve an interactive version of the notebook as a dashboard by activating the cuhvid conda environment and serving the Jupyter notebook using Voila by running the included conda_env.sh and serve.sh bash scripts:

source conda_env.sh  # may not work, see below
bash serve.sh

1.1 Problems activating the conda environment

You may not be able to activate the conda environment with the conda_env.sh script. If so, run:

# activate the project's conda environment
source /opt/anaconda3/bin/activate
conda activate cuhvid
# serve voila app
bash serve.sh

---

© Oliver Lambson, 2020