The release of the Risklayer’s interactive risk explorer platform was months away in our Development Roadmap, but we decided to go live with it today. We want to present data on the Novel Coronavirus (COVID-19) to help inform the public and decision makers of the current situation as soon as this data becomes available.
Virus tracking maps to help the fight against Corona
Sleep has become a luxury for our small team since we stepped in to fill the gap of reporting COVID-19 cases in Germany at the district level called Landkreise. Beginning on March 5, we posted maps of confirmed COVID-19 cases and updated these on Twitter daily. Here is a video of the daily progression of the number of people with confirmed infections throughout Germany from March 5th to March 20th. It turns out there is no open central reporting system in Germany, and the Robert Koch Institute which is the official government agency for reporting these values, can no longer keep up with the data, only reporting cases that are transmitted to them electronically. Even this data comes with a lag of 24-72 hours. In Germany there are 401 districts which report COVID-19 data individually on their websites. In addition, there is a laborious cross-verification process between these sources, and data reported by the 16 Health Ministries for each state (Bundesland), the Federal Health Ministry and the Robert Koch Institute.
The idea was to fill the gap until an official government reporting system with timely reporting of all confirmed cases steps in. But so far no one has stepped in, and the virus mapping effort has taken a life of its own. Noticing the traffic on Twitter, mapbox offered to help and is providing us free access to their mapping tools in order to help us reach more people with interactive maps. We started collaborating with various news agencies, such as the local Badische Neuste Nachrichten (BNN), and the Innovation Lab at Tagesspiegel as well as a number of other groups to crowdsource reported COVID-19 cases in Germany. What has surprised us most, is the collective intelligence process which has emerged, with many digital volunteers who followed us on Twitter, offering to help collect data and confirm the reports of COVID-19 cases across the various sources.
With the support of Map Box, we are now going live and releasing the interactive Risklayer Explorer platform to support the public and decision makers, with crowd-sourced interactive maps of the COVID-19 cases. The Risklayer Explorer platform will provide access to data on reported Coronavirus cases, deaths, recovered and infection rates on a daily basis.
As of midnight on March 20th, 18,257 confirmed COVID-19 cases and 48 deaths were reported by all districts in Germany (Figure 1). This is an increase of around 24% or around 3500 cases from the numbers reported 19th March, but a slight decrease in the general trend of over 30% increase per day that have been observed from March 5 onwards. It will still take several more days to understand if the suppression policies of staying at home and social distancing implemented across Germany on March 17 are taking effect. It is important to note that the number of confirmed cases of infected persons provide only a snapshot of the virus transmission from more than a week ago.
Observing the data coming in on a daily basis shows the measurable difference that testing makes. Here again, it’s important to caution that the confirmed cases that are seen provide a view only of symptomatic and suspected cases which are being tested, and not potential asymptomatic cases which can still spread the virus. To have a more true spread of the infection risk, a stochastic process random probability sample of Germany to find out where the virus is really hiding would provide a more true picture.
Figure 1 - Number of confirmed COVID-19 cases per district in Germany on 20.03.2020 24:00 GMT
Tracing the beginnings of Risklayer’s risk explorer platform
The risk explorer platform is the result of a collaboration between Risklayer GmbH and the Karlsruhe Institute of Technology’s Center for Disaster Risk Management and Risk Reduction Technology (CEDIM). It’s the continuation of an idea that started in 2011 when CEDIM embarked on a new style of disaster research called Near real-time Forensic Disaster Analysis to analyze large disasters immediately after their occurrence, assess their impacts, and retrace the temporal development of these events through an interdisciplinary team of scientists. This approach was pioneered by Professor Friedemann Wenzel at the Karlsruhe Institute of Technology, with the basic idea that technology developments and the immediate availability of open data lead to unprecedented opportunities for loss assessment of disaster events in near real-time. Crowd sourcing allows for rapid assessment of initial information on events. When combined with stochastic modelling methods and Artificial Intelligence (AI), the crowd-sourced data on hazard and impact parameters can be used to estimate potential damages and larger societal and economic impacts. Since 2011, CEDIM has been analyzing large disaster events, and providing various stakeholders from disaster response, to insurance and humanitarian aid with timely and science-based assessments of disaster impacts. Watch a TEDx talk on CEDIM’s Near real-time Forensic Disaster Analysis approach here by Dr. Bijan Khazai.
In 2015, Professor Wenzel and a team of scientists that worked closely together on CEDIM’s Near real-time Forensic Disaster Analysis research formed Risklayer as a spin-off out of the Karlsruhe Institute of Technology. From its founding, the goal at Risklayer has been to provide rapid analytics around disaster events to support decision makers with metrics and tools that measure impact, manage response and monitor the recovery from disasters. Risklayer’s approach in global stochastic and near real-time risk modeling is based on the PhD dissertation of Dr. James Daniell, for which he won one of only three KIT Doctoral Awards in 2016, and was recognized in 2017 as Natural Hazards Outstanding Early Career Scientist, by the European Geophysical Union, awarded to the leading Natural Hazards scientist under 35 in Europe. Risklayer’s innovative rapid loss estimation technology is linked to the global multi-peril damage and loss database, CATDAT, which is used to calibrate model outputs through high quality empirical data on historical losses. CATDAT has been continuously developed into the world’s largest and most detailed historical catastrophe loss database. The CATDAT database is sourced from over 35,000 original sources in over 90 languages, and has been cited in over 1000 newspapers, magazines and TV broadcasts worldwide.
Tracking COVID-19 on the Risk Explorer Platform
Dr. Andreas Schäfer, Chief Technology Officer at Risklayer and founder of CATNews, was actually planning to release Risklayer’s risk explorer platform in the autumn of 2020. The urgency of the global crisis we are facing with the Coronavirus pandemic, and the importance of providing timely data on the spread of COVID-19 across sub-national political boundaries to the public and decision makers has been fueling Andreas around the clock, to release the platform way ahead of schedule. With the support of Map Box, and the growing number of digital volunteers supporting our mapping activities, Risklayer will be releasing the platform free of charge with an immediate focus on tracking and analyzing the data on COVID-19 at the district scale in Germany.
Risklayer will also be compiling sub-national data on the spread of the Coronavirus across Europe and World-wide. Risklayer compiled it’s first sub-national COVID-19 map for the world on March 11 (see Figure). The Coronavirus infection data compiled from various official sources by Risklayer at the second-administrative level for all countries is unique. It shows the importance of scale and looking at the COVID-19 infection data in terms of population density (for example, looking at the data from March 11, the province of Lodi with 482 cases in Italy, has a higher percentage of the population infected (0.21%) than the highest infected population per total population at the state level in the Korean province of Daegu with 0.17%). This also points to the need for more granular travel advisories and identification of high-risk areas.
Figure 2 - Number of COVID-19 cases per million population for sub-national administrative districts (Admin Level 2) across the world on 11.03.2020 02:30 GMT
In the 9 days that have passed since this snapshot shown in Figure 2 on March 11, the global picture looks very different of course, with many more countries reporting infection cases. However, there is still lots of uncertainty in comparing these values. Maybe Mississippi is reporting no cases because it's not looking. Perhaps Zimbabwe reports zero cases because they don't have testing capability, not because they don't have the virus. All of this doesn’t account for the fact that tests are not conducted uniformly across the geography. The map of relative or absolute Coronavirus cases, convey at best where the tests are being intensively conducted. Even this assumes that the tests are conducted correctly, and that they are not being manipulated for political purposes. As the epidemiologist, Dr. Larry Brilliant, who helped defeat smallpox explains “we can beat the novel Coronavirus, but first we need a lot more testing”. He adds, “we need something that looks like a home pregnancy test, that you can do at home”.
Lots still to come with the Risklayer Explorer!
When the Risklayer’s risk explorer platform was envisioned, we did not have pandemics in mind. But we are responding to a need for tracking and analyzing reliable and granular data as it becomes available for an ongoing global catastrophe. This fits well with our mission of providing tools and analytics to rapidly track and analyze data as it becomes available following large or damaging catastrophes.
We believe what sets us apart from other catastrophe data aggregation services is that we provide a science-based lens on the interpretation of this data. In addition to the Coronavirus tracking tool just released, we currently also provide analytics on damaging earthquakes globally, including event characteristics, spatial footprints of the earthquake's intensity as well as the population affected by the earthquake. We expect to release many more features, and extend our global analytics to tropical cyclones, volcanos and extreme weather events within 2020. Learn more about our development roadmap on the Risklayer Explorer platform.
Corona tracking is offered free of charge
Risklayer does not make any money with its Corona-Tracking tool and hopes to be able to finance itself later with risk analytics for clients seeking its analytics services, for example operators of large hotel chains and wine producers worldwide – two sectors for which it is developing targeted tools and metrics. For example, through our analytics, we provide estimates of lost revenue to winemakers in disasters. Risklayer also supports Hotel Resilient, a scientific benchmarking and certification body for disaster risk management and climate change adaptation of hotels and resorts, through a dedicated Application Program Interface (API) for analyzing the risk of hotels and resorts to multiple perils, including the impacts of climate change.
We hope that you explore the Risklayer’s risk explorer site and find the data presented useful. If you find something not working or you have suggestions to help us improve the site, please don't hesitate to contact us!