Improbable simulation engineers contributing to the Rapid Assistance in Modelling the Pandemic (RAMP) initiative have successfully increased the speed of a pandemic transmission model by a factor of 10,000.
A month-long simulation of movement and viral spread involving nearly a million people can now be run in less than half a second. This allows experts to generate a huge amount of simulation data, to give insights into the possible outcomes of different policies and activities. Faster and more interactive modelling provides better insights and drives better science and better decision-making in response to complex, critical questions.
The RAMP initiative, convened by the Royal Society, brings together modelling expertise from a range of disciplines across academia and the private sector to inform the Government’s advisory bodies, including the Scientific Advisory Group on Emergencies (SAGE).
The group, comprising experts from Leeds, Exeter and Cambridge universities and University College London, is tasked with developing an agent-based model, where AI-driven representations of individuals, groups and areas interact, able to connect behavioural patterns with data on disease spread.
The result of this project is a detailed virtual model of Devon’s entire 800,000-person population, with Improbable integrating the output of the models into a visual interface, allowing for a clearer representation of the county’s movement patterns and the disease transmission levels they create, and the impact different behaviours and regulations may have.
This allows researchers and policy designers to visualise the highly complex information presented by the interaction of complex models, as well as better understand the consequences of different interventions and scenarios.
Improbable integrated disparate agent-based models into a single unified environment, using open source technology and Improbable’s expertise in managing multiple models and domains within synthetic environments.
The ability to run multiple simulations based on different premises and actions, and to project the results further into the future, means the model can now provide more information for policy designers experimenting with different policy options and intervention strategies, to explore the widest possible range of outcomes.
The model code has now been fully open sourced and integrated into the project’s codebase.
The simulation models the population of Devon. In this simulated view of Exeter, transmission risk is marked in red.
Joe Robinson, CEO of Improbable’s Defence and National Security business, said:
"Our team has enabled the creation of highly realistic virtual representations of the real world (Synthetic Environments) through the combination of multiple models from a range of experts. This interoperable solution delivers much greater speeds than seen previously. The faster your model runs, the easier it is to explore it interactively, and discover flaws that need improving, or surprising predictions that can inform decision-making at the regional or national level.
“This project used the excellent models built by the academic partners in the Urban Analytics project, but the learnings from the project could be shared to other regions of the UK, and we’re excited by the opportunities to help inform decision-making with the potential to save lives.”
Professor Mark Birkin, Programme Director for Urban Analytics at The Alan Turing Institute, said:
“The pandemic has demonstrated unequivocally the need to mobilise the best in mathematical modelling and data science to inform policy in the public interest.
“Massive enhancements in the power of our models through the collaboration with Improbable open up new horizons in the evaluation of multiple lockdown scenarios, assimilation of complex data relating to social and behavioural patterns and health outcomes, as well as opening the door to new academic insights in addition to enhanced decision-support.”
Professor Nick Malleson, Professor of Spatial Science at the School of Geography, University of Leeds, added:
“By speeding up the simulation so that it runs in seconds, rather than hours, Improbable has made it possible for us to conduct new research into the impacts of the pandemic that would otherwise not have been possible.
“Now that the simulation results can be generated almost immediately, we can test a much wider range of possible policy scenarios and quickly identify interesting features that should be explored further. This is much harder to do when we need to wait hours for the model to provide results.
“Also, when we start to scale up the model to simulate disease spread across all of England and ultimately the UK, we will be able to work with the simulation on our own computers rather than having to delegate the processing to bespoke high-performance computer grids. This is a much more flexible way to work and, again, will allow us to quickly explore different potential policies."
Founded in 2012 and headquartered in London, Improbable employs more than 700 staff, with the majority working in the United Kingdom.
Improbable’s games business provides better ways to make multiplayer games and helps multiplayer developers to meet any challenge. Improbable’s services include managed hosting & orchestration, networking, online services and development tools, as well as advice, support and full co-development. Improbable also makes innovative multiplayer titles using its own technology.
Improbable’s Defence and National Security business operates globally and combines its parent company’s software engineering experience with expertise in computational modelling, AI and data analytics. Its work focuses on adapting and extending Improbable’s multiplayer gaming technology to enable the most sophisticated military simulations and synthetic environments ever experienced.
For media inquiries:
Head of Global Communications
For business inquiries: