Can passively collected mobile-network data reveal useful public-health signals during an epidemic, without apps, contact tracing, or individual diagnosis? We investigate this question using mobility traces from more than two million mobile-network users in Madrid across two distinct phases of the COVID-19 pandemic. During the March 2020 lockdown, we identify hospitalization-related mobility patterns from sustained presence within hospital footprints and obtain a population-level estimate closely aligned with official hospitalization records. In June–July 2021, when restrictions and behavior had changed substantially, we detect prolonged home-stay patterns and compare them with Madrid’s official home-monitoring benchmark, while also testing whether recent mobility behavior can anticipate subsequent home-isolation behavior across more than 6.3 million user-week observations. Together, these analyses show that mobile-network data can capture useful public-health signals across different stages of an epidemic and provide a scalable source of information for public-health monitoring.

About Behafarid Hemmatpour

Behafarid Hemmatpour is a third-year PhD researcher at IMDEA Networks Institute (and UC3M), working with large-scale mobility data to build practical solutions for intelligent transportation and computational epidemiology. Her work spans spatiotemporal analytics and machine learning, including federated learning; the design of data-driven methods for smart street-parking management and urban traffic optimization, and predictive models that improve epidemic forecasting. She evaluates ideas through numerical simulations and real-world datasets, releasing reproducible, open-source code to support community use and extension. Before her PhD, Behafarid earned an MSc in Statistical Physics and Complex Systems at Shiraz University, where she investigated human mobility patterns and modeled cities as complex systems, work that seeded her current focus on scalable, data-driven methods for urban systems and public health.

Este evento se impartirá en inglés