/COVID-19-forecasting-dashboard

COVID-19 forecasting dashboard using data from John Hopkins University, PHE and www.gov.uk. An NHS Bed forecasting model for England is also added to the UK table and plotted. There is also a notebook for forecasting the Italy COVID-19 cases which implements a logistic model, and exponential model, and tries the fb prophet model. Important note: Please note that the case forecasting and bed forecasting model is experimental and not yet validated (as cases have not yet arrived since the model is being developed as the pandemic unfolds). The model is only as good as its assumptions and starting data both of which are likely to have limitations, especially this early in the pandemic and therefore it should not be used for clinical decision making.

Primary LanguageJupyter NotebookMIT LicenseMIT

COVID-19-forecasting-dashboard

This is a project which trys to track the COVID-19 outbreak and forecast the future cases. The notebook forecasting_dashboard.ipynb is a jupyter notebook which will try to forecast cases from yesterday's updated data from John Hopkins University as well as a few other sources such as www.gov.uk. It will run up a dashboard on local host to allow exploration of cases around the world as well as country case forecasting based on a simple logistic model. Further models will be added in future. The notebook UK_COVID-19-curve-fitting-free-bed-forecasting.ipynb is a notebook to delve deeper into the UK data, which aims to try and estimate the number of NHS beds currently free in different NHS trusts in England as well as forecast the number of beds which will be free in 1 week's time. It uses the Q1 2019 UK NHS bed data as a baseline before adding the COVID-19 hospitalisation forecasts and estimates on top. Important note: Please note that the case forecasting and bed forecasting model is experimental and not yet validated (as cases have not yet arrived since the model is being developed as the pandemic unfolds). The model is only as good as its assumptions and starting data both of which are likely to have limitations, especially this early in the pandemic and therefore it should not be used for clinical decision making.