Latent patterns of urban mixing in mobility analysis across five global cities

A team from MIT's Senseable City Lab, Harvard, and other universities published a groundbreaking AI study in Nature Cities, analyzing detailed travel surveys from over 200,000 residents across five major global cities: Boston, Chicago, Hong Kong, London, and São Paulo. The research leveraged a Graph Neural Network (GNN) to construct detailed spatio-temporal place networks for each city, feeding inputs of home-space, activity-space, and demographic attributes into a supervised autoencoder to predict individual exposure vectors. The core finding was that the structure of an individual's activity space—where they actually travel—explains most variations in place exposure, suggesting mobility shapes experienced social mixing more than sociodemographic characteristics, home environment, or transit proximity.

Key discoveries include a significant 16% underestimation of social mixing levels when inferring socioeconomic status from residential neighborhoods compared to using self-reported survey data. The study provided data-driven support for the 'second youth' hypothesis, showing individuals over age 66 experience greater social mixing than those aged 55-65. Conversely, teenagers and women with caregiving responsibilities exhibited lower mixing levels. An important nuance emerged from ablation tests: while different income groups may experience similar quantitative levels of mixing, their activity spaces remain stratified by income, resulting in structurally different qualitative social mixing experiences. The research also found that proximity to major transit stations reduces the influence of individual socioeconomic status on social mixing patterns across all five cities.