Innsikt

When can we travel safely again in Europe?

Interactive Visual Analysis is a powerful technique for investigating relationships between different data. We have been trying out Enlighten-web in our analysis of COVID-19 infections.

Sist oppdatert: 3. jul. 2020
Publisert 3. jul. 2020
Southwold Beach huts 2
cjtrigg / CC BY (https://creativecommons.org/licenses/by/2.0)

We have, since the outbreak of SARS-CoV-2 began, been looking at trends in all countries worldwide using the tool developed at NORCE for Interactive Visual Analysis. The data is released on a daily basis from the European Center for Disease Prevention and Control (https://www.ecdc.europa.eu/en) and we retrieve the latest figures once per day to display in our Covid-19 dashboard. Note - this is the only data source we are using in our analysis, and we cannot take account of how much testing is carried out or whether the number of cases for each country is correctly represented.

The analysis and figures in this article are all carried out using our Enlighten-web tool. Enlighten-web is a powerful web application and has been developed through a number of research projects at NORCE. Graphical display relies heavily on GPU processing. In this way we can display millions of points in the scatter plots that you see.

In this insight article we illustrate a number of effects of imposing and lifting lockdown constraints during the SARS-CoV-2 pandemic. An example is the current apparent new wave of cases for the United States (as of 30th June). The number of cases is increasing dramatically. We can plot the running number of total cases but it doesn't give us much insight into what is happening.

Covid19 fig1
Figure 1: Total COVID-19 cases for USA (as of June 30th). The green bar in the bar chart shows which country is selected. The points for the same country are then highlighted on the right. The grey points are all the other non-selected countries.

With our dashboard we have also calculated the second derivative - ie the acceleration or deceleration of the number of cases. This tells us a little bit more about what is going on. We see for example that back on 26th March the outbreak was at it's peak for the USA. A new peak has now formed with the top at 21st June and it seems - for now - to be coming back down again. The number of daily cases will continue to grow though as long as that second derivative is above zero.

Covid19 fig2
Figure 2: Second derivative of the running total number of cases for the USA

The authorities in Norway have announced that from the 15th July it may be possible to travel outside of Norway to countries in Europe without quarantine on return if the countries meet certain infection criteria. We have been using Interactive Visual Analysis to try to determine what countries fall within one of the criteria which is that there should be less than 20 registered cases per 100 000 of population. We have to assume that they mean how many are currently sick. We have the daily worldwide case reports from ECDC, but that just tells us how many are reported sick each day, not how many are currently sick. To get the total number of cases, ie how many are currently sick, we have to make an assumption about how long someone is sick for and the general impression seems to be that once infected, someone might be sick for 14 days or longer. In our calculation we take the rolling sum of the daily cases for 14 days to get an estimate of the actual number of cases per 100 000 of population.

In the figure below the bar chart on the right is used to make a selection of the continent (Europe) and we see on the top left the 9 countries in Europe with the highest active cases per 100 000 population

Covid19 fig3
Figure 3: Selecting Europe as the continent by clicking in the par chart on the right shows data for all countries in Europe simultaneously. Multiple countries are shown highlighted on the lower left using a colour scale to separate them. The bar-chart on the left shows the selected countries ranked from highest to lowest according to the number of cases per 100 000 population.

We can also use a map to select the countries that we might want to travel to. The example below shows a geographic selection of northern Europe from Ireland in the west to Sweden and Czech Republic in the east. By far the "worst" country in this region is Sweden with over 140 Covid-19 cases per 100 000 population with Luxembourg and the United Kingdom in second and third place. We can also click through each country in turn to see the historical development over time.

Scatter plots in Enlighten-web implement "brushing and linking". This means that when we make a selection in one plot, the same data is also highlighted in the other plots on the same page. In the figure below we first select some countries in northern Europe using the map, and then interactively select values in the scatter plot that are higher than the criteria we are looking for - ie 20 cases per 100 000 population. The bar chart on the same page will then show the countries that currently don't meet the criteria set by the Norwegian authorities for safe travel.

All of our analysis is based on our existing Enlighten Web analysis tool. Please be aware that the infection data is updated daily and the interactive figures may be different when you read this article. Some intermittent negative values and trends may also be seen in the case when authorities adjust the national figures and a negative number of cases are included in the data to compensate.

The demonstration is available at this location where you can try it for yourself. Hopefully it will work for you, but if not then please get in touch.

Aktuelt

NORCE LNLYR High Res 01
17. sep. 2019

Verdens første passasjerfly for miljøovervåking er norsk