COVID-19 is one of the fast-spreading diseases the world has ever seen, with the most recent Omicron variant now considered to be as contagious as the previous frontrunner, measles.
But predicting the spread of the virus now, and any variants to come, relies on complex epidemiological modelling, which can consider different scenarios.
Using an agent-based modelling framework, and the computing power of Pawsey’s supercomputers, researchers from the Department of Computer Science and Software Engineering at the University of Western Australia (UWA), has now developed a first of its kind epidemiological model to better shape the health care system’s plan to combat COVID-19.
The first reports of epidemiological modelling date back to the eighteenth century, when Swiss mathematician Daniel Bernoulli developed a mathematical model to calculate the increased life expectancy of those inoculated against smallpox.
Epidemiological modelling plays a significant role in influencing decisions as authorities handle the spread of an infectious disease, such as COVID-19.
To develop a model fit for the pandemic, UWA’s Department of Computer Science and Software Engineering collaborated with the Institute for Healthcare Informatics and the Department of Physics at the State University of New York (SUNY).
The goal was to create a model able to predict future trends in hospitalisation and admittance to intensive care due to the spread of COVID-19. It was the first time that an agent-based modelling framework had been used to frame COVID-19 projections for local communities and cities.
Agent-based modelling is a computational model for simulating the actions and interactions of autonomous decision-makers, to understand the behaviour of a system and what governs its outcomes.
The team at UWA and SUNY developed a simulation to closely reflect the spread of COVID-19 in New York State in the early days of the pandemic. They chose two types of entities for this model — home groups and public places — and started each simulation with 0.01 per cent of the population becoming infected.
The simulations were run on two of the supercomputers at Pawsey Supercomputing Research Centre, Magnus and Zeus.
Using the supercomputers, the team was able to compare the results of their simulations to actual data taken from New York State, which was seriously impacted by COVID-19 in both 2020 and 2021.
By comparing the model to New York State’s actual COVID-19 figures, UWA and SUNY were able to identify where the accuracy of their model diverged. Now, it is expected that once the necessary parameters are extracted for a given region and the population involved, this model can predict future trends in hospitalisation and admittance to ICU.
With that information in hand, local communities and cities can better prepare for future demands on health care and identify how to provide services while keeping the economy from griding to a standstill.
You can read the paper published in Physica A: Statistical Mechanics and its Applications in January 2022 by visiting this link