The analysis of extensive datasets on human mobility patterns reveals the existence of two distinct classes of individual travel behaviour, returners and explorers, reports a study published in Nature Communications. This observation allows the creation of a new model of mobility which may help contain the spread of epidemics. Variability in human movement is accompanied by a rather high degree of statistical predictability of global patterns at the population level. The increasing availability of tracking datasets has now made it possible to develop and verify quantitative mobility models.
The study was conducted by Luca Pappalardo and Dino Pedreschi (Università di Pisa), Fosca Giannotti and Salvo Rinzivillo (Isti-CNR Pisa), Albert-Laszlo Barabasi (Northeastern University Boston) and Filippo Simini (Bristol University).
The researchers analysed an anonymized dataset composed of 67,000 individuals' call records over a three-month period, and compared it with the GPS traces of roughly 46,000 vehicles travelling through central Italy over a month. For each tracked person, they compared the overall distance travelled to their recurrent mobility, obtained by looking at their most visited locations. The researchers found that the mobility behaviour of a subset of individuals, the explorers, could not be approximated by their recurrent trips, as opposed to the well-defined patterns of the returners. Since existing mobility models could not account for this difference, the research team developed a new model to reproduce their findings through simulated travel patterns.
Explorers, by virtue of their non-recurring mobility, were found to influence the potential spread of epidemics. In fact, simulated global mobility networks for the population of Tuscany show that, with an increasing proportion of explorers, the chances of a widespread epidemic also increase. This suggests that improved understanding and modelling of human mobility have the potential to help predict the future spread of disease.
Figure: A fragment of the GPS trajectories produced by 150,000 vehicles traveling in the metropolitan area of Pisa (blue) and Florence (red) during one month. This geo-referenced visualization of the data demonstrates the ability of Big Data to portray social complexity.