A new model of human movement may help cities plan better transportation, respond to pandemics, and manage natural disasters. Researchers at Rensselaer Polytechnic Institute built the model using anonymized GPS data from roughly 2 million cellphone users in the United States, tracked over six months. Their findings were published in the Journal of the Royal Society Interface.
According to Phys.org, most existing models of human mobility assume people follow similar movement rules regardless of how far they travel or where they are going. The new research challenges that assumption.
The senior author of the paper, Jianxi Gao, said the idea came from an unexpected place. "The original idea came from a sermon at my church, where a visiting pastor showed us the picture View of the World from 9th Avenue," Gao told Phys.org. "In this image, the area close to Ninth Avenue appears very large, while places farther away are compressed into a much smaller space. This led me to ask whether this was simply artistic perspective or whether it also reflected how people perceive space and time."
That question drove the research team to look at how people perceive space not just in terms of physical distance, but also in terms of their personal connections and relationships to specific places. The insight shaped how they built their model.
To develop it, the team drew from ecology and biology. Animals switch between different movement strategies depending on what they are trying to accomplish, and the researchers applied that same logic to human travel. The result is what they call a mode-switching model of human mobility.
Gao described why earlier approaches fell short. "Previous mobility models usually focused on travel patterns between pairs of locations (for example, number of trips between two locations), but they often failed to reproduce network-level properties such as modularity and average shortest-path length," he said.
The dataset the team used was large. One portion came from six months of privacy-enhanced GPS trajectories from approximately 2 million anonymized users in the United States. A second dataset added further scale to the analysis, though details on that set were not fully available in the published summary.
Gao also placed the new paper in context with earlier work from his team. "This is a sister paper to our earlier study, in which we summarized the main discoveries about a universal inflation law of human mobility," he said. "The present paper takes the next step by asking what underlying mechanism could generate the inflation law."
The inflation law refers to a pattern the team had previously identified in how people spread their movements across space, with nearby locations receiving far more visits than distant ones, in a pattern that stretches and compresses like a map drawn from a single vantage point.
Understanding those patterns at a deeper level has practical stakes. Engineers and city planners use mobility models to improve transportation systems and design more efficient urban environments. Public health officials rely on similar models to track how diseases spread during outbreaks. Emergency managers use them to anticipate where populations will move during disasters.
The new model, because it accounts for how people perceive and relate to places rather than just how far away those places are, may produce more accurate predictions in all of those areas. The researchers have not yet announced follow-up studies, but the paper represents a step toward mobility models that more closely reflect how people actually decide where to go.
