Ships are an important means of transportation for both people and goods, which also makes them potential hotspots for diseases, especially those with a longer incubation period. The recent clusters of COVID-19 infections on cruise ships such as The Diamond Princess and the Grand Princess are cases in point, but are certainly not the only examples. Based on previous research we conducted with vessel tracking data, we know that this data is a valuable source of information. Therefore, we set out to explore whether insights from this type of big data could be included in epidemic modelling of diseases to inform a more efficient and timely operational response.
Although passenger ships have nowadays largely been replaced by airlines, they still represent a substantial amount of traffic in island nations, and they continue to play an important role in international trade and tourism. Did you know for example that in 2018, cruise ships transported some 28.2 million passengers worldwide?
Automatic Identification System (AIS) data forms a global database of maritime traffic. Most large commercial, international, and passenger ships must be equipped with an AIS transmitter, which reports dynamic details about the ship’s position – such as its location, speed, and course over ground – and static details, including a ship’s identifier, type, and flag. This information is routinely used to monitor port security and detect fishing behaviour. More recently, it has also been used to study search and rescue operations at sea.
Check out our project on Using Big Data to Study Rescue Patterns in the Mediterranean
The current standard for calculating disease import risk at the international level is epidemic modelling that uses flight network data on commercial airline routes and capacity. However, vessels can also be global carriers for infected people and disease vectors.
We sought to understand how AIS traffic data can inform epidemic modelling using two case studies. The first looked at the plague outbreak that hit Madagascar in 2017, affecting an estimated 2,348 individuals and resulting in 202 deaths. The second study analysed the current pandemic caused by the novel coronavirus known as COVID-19. Our methodology consisted of gathering information from ports and ships and constructing an origin-destination matrix.
We learned that we can use AIS data to visualize individual ship trajectories to understand, for example, whether ships came to shore in cities with high risk of infections, and to account for their arrival times in future ports in near real-time. The study also showed that there is a widespread reach of ships travelling between islands, in our case with vessels reaching three or more continents within a one month period.
This type of information could be useful in complementing insights from flight data, especially when modelling the spread of diseases in port cities and island nations. We hope our research can inspire future work on how combining vessel tracking data and flight network data could change the outcome of disease simulations and risk estimates.