Dustin CARLINO, Alan Turing Institute, United Kingdom
Anna ZANCHETTA, Alan Turing Institute, United Kingdom
Hadrien SALAT, Alan Turing Institute, United Kingdom
Fernando BENITEZ, Alan Turing Institute, United Kingdom
Mark BIRKIN, Alan Turing Institute, United Kingdom
Several studies are investigating the use of non-pharmaceutical interventions (NPIs) against the COVID-19 pandemic (Desvars-Larrive 2020). As governments are seeking the proper combination of national and local policy measures, new methods for understanding NPIs and the effects of lifting lockdown restrictions are at the centre of academic discussions (Haug 2020). This project introduces Dynamic Micro-simulation Model for Epidemics (DyME), a multidisciplinary approach to studying NPIs, combining epidemiological modelling and urban analytics. It was developed under the UK Royal Society's Rapid Assistance in Modelling the Pandemic (RAMP) initiative, a collaboration between multiple British universities and external partners, led by the Leeds Institute for Data Analytics. This talk describes the process of scaling it to the national level for England, its implications, and our vision on how this approach can contribute to a better urban decision making process.
The DyME is an agent-based model that combines SPENSER (Lomax 2017), QUANT (Batty 2021), and other data sources. A synthetic population is created, with each individual bearing appropriate demographic, health, and employment attributes, and a set of daily activities. The DyME model simulates these people visiting shared locations. Initial infections then spread based on an updated Susceptible, Exposed, Infected and Recovered - SEIR transmission model.
This model was recently evaluated with a case study in the county of Devon in England (Spooner 2021). Our work is to scale this approach to the entire UK defining “what-if” scenarios for more applied urban analytics using spatial microsimulation. The software can help guide policy decisions around NPIs. As such, future work will explore the effects of multiple COVID variants, vaccinations, more realistic travel behaviour on weekends, and transmission through car-pooling and public transit.
Mots clés : covid|agent-based modeling|spatial model|synthetic population|gis
A105245DC