Tyrone Burke, May 1, 2020
Photo credit: Luther Caverly
Tracking the Spread: Predicting COVID-19 Transmission
The amount of time we spend puzzling over exponential growth curves has increased, well, exponentially.
But the COVID-19 models shown at daily press briefings have their limits. As we move from containing the virus to managing it, other advanced modelling techniques could help identify outbreaks before they occur, and limit the damage they cause.
“Most COVID-19 models have been built to answer a specific question — how do we flatten the curve? This is a major problem we need to attack, because if we don’t control the number of cases, the health care system will be overwhelmed,” says Dr. Gabriel Wainer, Professor of Systems and Computer Engineering and Carleton’s Associate Chair for Graduate Studies.
These models aim to predict how many people will need treatment, and whether our hospitals will be able to manage the case load. They are updated daily, and adjusted as additional data becomes available.
“But as soon as you get to the real world, everything that was completed yesterday will need to be adjusted to reflect new data” Wainer says.
“Models predict the spread of the pandemic based on certain conditions and assumptions. If they are not met, the model needs to be changed. For instance, if outbreaks in retirement homes are not properly handled, or people don’t follow the lockdown guidelines that the models assume, you will need to change the model and run new simulations And if you obtain better from additional testing, you might need to work backwards to understand what has been happening.”
Wainer is making models that will use detailed information from Building Information Models (BIM) to predict how the virus circulates within large buildings — and predict who could have been exposed.
“One recent study that has been discussed in several news outlets showed that at a restaurant in Guangzhou, China, three family clusters were infected with COVID-19 by droplet transmission that was influenced by the airflow from air conditioning units. The droplets had travelled farther than expected,” says Wainer.
Prior to the beginning of the pandemic, Wainer’s research focused on modelling building use to maximize energy efficiency. He’s the lead researcher for the SUSTAIN project – short for Sensor-based Unified Simulation Techniques for Advanced In-Building Networks.
On SUSTAIN, Wainer works with Dr. Stephen Fai (center) and Dr. Liam O’Brien (left). The project uses 3D BIM models to create digital replicas of buildings. The BIM information is combined with simulation tools to detect a building’s occupancy and study how it is being used. Sensors identify the presence of people in different areas of a building, which allows building managers to optimize energy use. Heating, cooling, and lighting can all be reduced when they aren’t needed.
But COVID-19 created an entirely new application for this research.
“Now that there are people that might be sick in these buildings, we can change the simulation models fairly quickly. We can move from studying energy efficiency based on how many people are working in a given building, to identifying how many people could have been infected with coronavirus there, if two people have tested positive,” says Wainer.
“We can model where the virus might have travelled inside the building. If there is a wall, the virus cannot go through it, so we put that into the model. If an infected person sneezes, the virus could be carried through the air conditioning system.”
New modelling techniques could identify areas where COVID-19 exposure could have occurred, and help deploy limited testing resources to people who are most likely to have been infected with the virus. Wainer is currently focused on buildings on-campus, but the models that his team is developing could be applied in off-campus buildings too. Even in a lockdown, there are millions of Canadians who use large commercial buildings and condominium towers every day.
The COVID-19 situation is unfolding rapidly, and there’s no guarantee that the new models will be ready in time to fight this pandemic. But models like these could help ensure that we are better prepared the next time an unknown illness threatens our way of life.
“This research is just starting, but we have been reading about pandemics for a few years,” says Wainer.
“If it is not coronavirus, it could be avian flu, ebola, or something even worse than that. We are closer to wildlife due to the expansion of cities, industrial farming, forestry, mining and other industries that affect the environment. There are millions of viruses, and with the way that the world has globalized, those viruses will spread. It is a major issue that we will face in the coming years, and we will need advanced tools like this to help manage people as they come back to normal life.”
Share: Twitter, Facebook