Machine learning-based short-term forecasting of COVID-19 hospital admissions using routine hospital patient data

This repository contains the code of the analysis of the paper "Machine learning-based short-term forecasting of COVID-19 hospital admissions using routine hospital patient data" by Martin Wohlfender et al. A preprint is available on medRxiv.

Licences

• The code stored in this repository is available for download and use under the GNU Affero General Public License Version 3.
• Electronic health records data has been obtained from the Insel Data Science Center IDSC. Some of this data is made available in this repository in aggregated and anonymized form for download and use under the creative commons license CC BY 4.0.
• Wastewater data has been retrieved from the Swiss Federal Institute of Aquatic Science and Technology eawag under the creative commons license CC BY 4.0.

Overview of content of repository

• The aim of this repository is to provide all necessary code (written in R and Python) to reproduce the statistical analysis of the paper cited above.
• The whole R code is structured in an R-project (hospital_admission_forecasting.Rproj).
• Before running any other R file, the file setup.R (contained in folder R) needs to be run. In this file, all paths to data and results files are defined (with respect to the path of hospital_admission_forecasting.Rproj).
• R files are grouped by topic (data processing, creating plots, ...).
• All models except last observation carried forward and linear regression were run on the high performance computing cluster of the University of Bern, UBELIX.