Forecast and visualization of the spread of SARS-CoV-2 at regional and local level

In the spring of 2020, SARS-CoV-2, the virus that causes COVID-19, spread around the world. Healthcare professionals and researchers have so far observed that the spread of infection often manifests itself in the form of local outbreaks, where an increased number of individuals living together in the same residential area develop COVID-19-related symptoms.

In order to prevent and mitigate such local outbreaks, also known as cluster outbreaks, it is essential that all those who develop COVID-19-related symptoms have the possibility to get tested as soon as possible so that contact tracing and isolation can be initiated. It is, nevertheless, extremely challenging to detect local outbreaks at an early stage. Meanwhile, disease surveillance using different metrics may enhance early detection and enable early interventions.

Informed by this challenge, the CRUSH Covid research project was created to find new ways of early detection of signals that indicate an increased infection transmission. Results are communicated weekly to local health authorities to provide the basis for directed, tailor-made interventions at a local level.

Background CRUSH

CRUSH Covid is an innovative, multidisciplinary research project which aims to detect, map and prevent local COVID-19 outbreaks at an early stage. The ultimate goal is to assist health authorities to curb local outbreaks of COVID-19 within Uppsala County, as well as to inform individuals about the status of the pandemic in their neighborhood. In order to do so, the Uppsala Region has joined forces with prominent researchers from five distinct academic departments of Uppsala University, including general medicine, molecular and medical epidemiology, information technology, environmental microbiology, statistics and media and communication science. Examples of data used include, but are not limited to virus measurements in waste water samples, testing rates for COVID-19, available information concerning infection tracing, anonymized inputs from the helpline 1177 Vårdguiden, 112 ambulance calls and Google Mobility statistics, among others.

By using such a diversity of data sources, CRUSH Covid continuously monitors temporal trends and sheds light over key spatial developments in an attempt to timely identify signs of infection before the onset of an exponential spread of the virus. A comprehensive and dynamic report is then generated on a weekly basis, contributing to decision-making regarding sampling strategies and other targeted interventions as initiated by the Infection Control Unit of Uppsala Region (Smittskyddsenheten). As soon as a local outbreak is identified, a mobile sampling unit may, for instance, be dispatched to the inflicted neighborhood to offer and encourage testing, as well as to assist with infection tracing on site.

Join Hack the Pandemic here

The Challenge

Main challenge

Forecast the spread of the virus ahead of time by formulating a prediction model valid for one or two weeks

from now. The spread of the virus might be reflected on a national (Sweden), regional (counties within Sweden) or local (municipalities within Sweden or Uppsala Län) level. You may use test positivity rates (number of positive tests out of all tests) or the notification rate (number of cases per capita) as your primary outcome.

BONUS challenge

We would like you to think about alternative ways of effectively communicating the spread of the virus to the general public. For example, what visualization techniques would you use or would you like to see implemented on the CRUSH Covid dashboard?

Here you will find some useful links with readily-available datasets that you may use for the purpose of this challenge. Feel free to take a creative approach and use alternative data sources that you may consider of importance to better predict the spread of infection.

Join Hack the Pandemic here