There is science of infectious disease transmission hidden in a day's route of commuting from Seoul, passing through Gyeonggi and Incheon, and returning to Seoul. A domestic research team created a computer model of complex commuting routes and reproduced how infectious diseases actually spread.
Lee Jae-woo's physics research team at Inha University said on the 15th that it developed a new epidemic spread prediction model that reflects people's travel routes using domestic mobile carrier data, the "commuting metapopulation model (CMPM)," and published it in Chaos, an international journal issued by the American Institute of Physics (AIP).
The "standard metapopulation model (SMPM)" used in ecology and infectious disease research divides the population into "single blocks at the city level" and assumes that everyone mixes similarly within the same area. This approach is simple and fast to compute, but it does not capture the movements of people who commute to different areas every day in real life.
For example, existing models cannot distinguish between people who work in Seoul but live in Gyeonggi Province and those who live in Seoul. As a result, they produce outcomes different from reality, such as diseases spreading too quickly or regional differences disappearing.
Lee grouped the population not only by city units but also by commuting routes. The model directly incorporates the real-life pattern of contacting different people at workplaces during the day and near home at night. This allows for a more precise analysis of which routes transmission travels through.
The team used 2019–2020 communication data from KT, the country's No. 2 mobile carrier, to run simulations of the spread of the novel coronavirus disease (COVID-19). As a result, areas with high population density and large commuting populations, such as Jongno District in Seoul, saw rapid transmission that quickly spread to surrounding cities. In contrast, areas with fewer external connections, such as Gangneung or Jeju City, saw transmission spread much more slowly. Existing models barely reflected these differences.
The team also examined how the speed of spread and the timing of the peak vary depending on where the outbreak begins. The commuting metapopulation model showed a wide range for the peak, from the 83rd to the 119th day, depending on regional commuting characteristics. Existing models mostly predicted nearly the same timing, around the 100th day.
The researchers expected that the commuting metapopulation model would be useful for predicting infectious disease spread in countries with complex interregional movement. Lee said, "Rather than uniform lockdown measures, the government and disease control authorities could devise tailored response strategies for areas with large commuting populations and vulnerable areas with little external consolidation."
By comparing telecommunication data with randomly generated virtual data (a random model), the team also confirmed that the commuting metapopulation model based on real-world data reproduces far more realistic patterns of spread. They explained that actual mobility data contain structural features—such as directions of movement and patterns of people concentrating in certain areas—that cannot be discerned from simple statistics alone.
Lee Jae-woo said, "Everyday travel routes are not just simple lifestyle patterns; they are key factors that determine the flow of infectious disease spread," adding, "Our goal is to help disease control authorities devise more precise and efficient countermeasures by analyzing in real time how commuting patterns affect the spread of infectious diseases using the commuting metapopulation model."
References
Chaos (2025), DOI: https://doi.org/10.1063/5.0284992