The increasing agglomeration of people in dense urban areas coupled with the existence of efficient modes of transportation connecting such centers, make cities particularly vulnerable to the spread of epidemics. Here we develop a data-driven approach combines with a meta-population modeling to capture the interplay between population density, mobility and epidemic spreading. We study 163 cities, chosen from four different continents, and report a global trend where the epidemic risk induced by human mobility increases consistently in those cities where mobility flows are predominantly between high population density centers. We apply our framework to the spread of SARS-CoV-2 in the United States, providing a plausible explanation for the observed heterogeneity in the spreading process across cities. Based on this insight, we propose realistic mitigation strategies (less severe than lockdowns), based on modifying the mobility in cities. Our results suggest that an optimal control strategy involves an asymmetric policy that restricts flows entering the most vulnerable areas but allowing residents to continue their usual mobility patterns.

人口密集的城市地区日益集聚，加上连接这些中心的高效交通方式的存在，使城市特别容易受到流行病的传播。在这里，我们开发了一种数据驱动的方法，结合元人口模型来捕捉人口密度、流动性和流行病传播之间的相互作用。我们研究了从四大洲中选出的 163 个城市，并报告了一个全球趋势，即在人口流动主要在高人口密度中心之间流动的城市中，由人口流动引起的流行病风险持续增加。我们将我们的框架应用于 SARS-CoV-2 在美国的传播，为观察到的跨城市传播过程的异质性提供了合理的解释。基于这种洞察力，我们基于改变城市的流动性提出了现实的缓解策略（不如封锁严重）。我们的结果表明，最佳控制策略涉及一种不对称政策，该政策限制进入最脆弱地区的流量，但允许居民继续其通常的流动模式。