2019 Novel Coronavirus Global Risk Assessment
As of , total cases have been confirmed. Of these, there are confirmed cases in mainland China and confirmed cases in other countries.
This site compiles results obtained by a computational / mathematical model for the expected global spread of the novel coronavirus that originated in Hubei Province in China in December 2019. The model's foundation is the worldwide air-transporation network (WAN) that connects approx. 4000 airports with more than 25000 direct connections.
The network theoretic model is based on the concept of effective distance and is an extension of a model introduced in 2013 in the paper The Hidden Geometry of Complex, Network-Driven Contagion Phenomena, D. Brockmann & D. Helbing, Science: 342, 1337-1342 (2013).
Global risk assessment - results in a nutshell
The relative import risk by country is shown below by percentage. The concept of relative import risk is explained in more depth below.
Update Feb. 4th, 2020:
The computed relative import risk numbers account for the current situation in Mainland China. The closure of the airport in Wuhan and the distribution of confirmed cases in the country is factored in.
Figure 1: Relative import risk in the top 30 countries (excluding mainland China). The inset depicts the top airports for a selected country. Hover over individual countries to visualize airport data for that country. Countries with reported confirmed cases of 2019-nCoV infections are marked in red and current number of confirmed cases is listed on the right. More information and more refined analyses are provided in the Import Risk Analysis section. The national import risk is the cumulative import risk of all airports in the respective country.
Background
The current outbreak of the 2019-nCoV virus started in Wuhan City, Hubei Province, China. While the first cases were reported as early as December 8, 2019, the outbreak gained global attention on December 31, 2019, when the World Health Organization (WHO) was alerted to “several cases of pneumonia” by an unknown virus.
The new virus was soon identified as a novel coronavirus and named 2019-nCOV. It belongs to the family of viruses that include the common cold and viruses such as SARS and MERS. On January 20, 2020, it was confirmed that the coronavirus can be transmitted between humans, greatly increasing the risk of a global spread.
Current situation
The spread of the virus outside of Wuhan is dominated by air travel. Wuhan, the seventh largest city in China with 11 million residents, is a major domestic air transportation hub with many connecting international flights. While the city has been effectively quarantined on January 23, 2020, the virus has already spread to other Chinese provinces as well as other countries.
As of , total cases have been confirmed. Of these, there are confirmed cases in mainland China and confirmed cases in other countries.
A real-time dashboard of cases is provided by Johns Hopkins University.
Modelling the spread of 2019-nCoV
The model estimates the:
- Relative import risk
- Most probable spreading routes
- Relative arrival time
for airports, countries, and continental regions worldwide. Details are provided in the model section.
Relative import risk
What is relative import risk ?
By looking at air travel passenger numbers, we can estimate how likely it is that the virus spreads to other areas. The busier a flight route, the more probable it is that an infected passenger travels this route. Using these probabilistic concepts, we calculate the relative import risk to other airports. When calculating the import risk, we also take into account connecting flights and travel routes that involve multiple destintions.
Probabalistic concepts such as import probability, invasion rate, and both relative and absolute import risk are frequently misinterpreted and mistaken for each other.
Therefore, in order to understand the data presented on this site, it is necessary to understand the distinction between these concepts. The relative import risk is defined in the following way:
If an infected individual boards a plane at airport A in an affected region, the relative import risk P(B|A) at airport B quantifies the probability that airport B is the final destination for that individual (irrespective of non-direct transit routes).
Say, 1000 infected individuals board planes at Wuhan Tianhe International Airport (WUH). An import risk of 0.208% at Paris Charles de Gaulle International Airport (CDG) means that, of those 1000 individuals, only 2 are expected to have Paris as their final destination.
Figure 2: Relative import risk explained. Say, 100 infected passengers board planes at airport X, with destinations at other airports in the network, potentially going through transit airports in the process, e.g. the passengers travelling to D and F in the illustration above. The relative import risk at a chosen destination is simply the fraction of the 100 individuals that entered at X and ended up at the destination airport. In the figure, transit passengers are depicted in grey. Note that the transit airport E has an import risk of 10% because 10 individuals have E as a final destination, even though 40% of the passengers that entered at X went through E en route to a different destination.
Most probable spreading routes
Given an outbreak location and an origin airport close to it, the model identifies the most probable spreading routes to all other airports in the network. Even though passengers can take different routes to a final destination, in the worldwide air transportation network, global spreading patterns are dominated by the most probable paths. When viewed from the origin airport (the root node), these paths make up a shortest path tree that shows how the spreading process can reach all other airports in the network. These trees are important for identifying what airports play a pivotal role in distributing the spreading process thoughout the network.
An interactive visualization of the most probable routes and effective distances is offered in the Route Analysis & Effective Distance section.
Figure 3: Most probable path tree rooted at Wuhan Aiport (WUH). The bottom node represents WUH. All other airports in the network are connected by the most probable paths to the destination. Airports with a large number of branches play an important role in distributing the spread across the network.
Effective distance and expected arrival time
In the figure above, the vertical distance to the root node represents the effective distance to the outbreak location. The notion of effective distance was derived from traffic flux statistics and introduced in the paper The Hidden Geometry of Complex, Network-Driven Contagion Phenomena, D. Brockmann & D. Helbing, Science: 342, 1337-1342 (2013). In a nutshell, two airports are effectively close to one another if there is substantial air travel between them.
Effective distance has been shown to be a much better predictor of expected arrival time of an epidemic than traditional geodesic distances.