image: Folashade Agusto employed computer modeling and large datasets to better understand how COVID-19 was transmitted in one community in South Africa during the worldwide pandemic.
Credit: Wikimedia Commons
LAWRENCE — University of Kansas researcher Folashade Agusto trained as an applied mathematician, though today she’s an associate professor of ecology & evolutionary biology.
She uses that mathematical training and computers to model infectious diseases. Her goal is “to identify ways in which we can mitigate the risk they pose to humans,” she said. “But I also do models involving animals as needed — and in recent times, I've started looking at plants as well.”
Recently, Agusto led research appearing in the peer-reviewed journal PLOS One that employs computer modeling and large datasets to better understand how COVID-19 was transmitted in one community in South Africa during the course of the worldwide pandemic. She wanted to understand how lockdowns, restrictions and human behavior affected the course of the disease.
“With COVID, we were interested in looking at how policies and human behavior affect transmission,” she said. “The community we focused on in South Africa is Gauteng.”
The KU researcher based her computer model on South African census data on population density and household sizes from the Gauteng province, epidemiological data from COVID studies and timelines of government policies in the city intended to curb the spread of the disease.
“This was from March 2020, when governments around the world announced sweeping lockdowns at different points in March, as they did here in the United States,” Agusto said. “Globally, there were lockdowns, but they varied by policy. We wanted to see what the levels of household infection and outside infection were during those periods, given the different policies being implemented.”
Rather than modeling each individual — because of computing power and time constraints — the KU researcher and her colleagues divided the population into four density groups.
“We had access to both population and household size data for the four groups,” Agusto said. “P1 represented low density, then P2, P3 and P4, with P4 being the most densely populated areas. We also looked at household sizes: one-person households, 2-3 people, 4-5 people and six or more. The dataset also included a zero-household category, which we assumed represented people experiencing homelessness. From this, we could analyze how density and household size affected COVID transmission patterns.”
Moreover, Agusto employed “agent-based” computer modeling, an alternate approach to COVID computer models that use differential equations to examine populations as a whole, then segment people according to the stages of infection.
“By contrast, the agent-based model is computer-based and focuses on individuals,” Agusto said. “It uses ‘if-then-else’ rules to simulate how a person goes about their daily life and makes decisions. These rules are also tied to probability. If someone does a certain action, what happens next? It’s a more natural way to encode human behavior compared with differential equations. Traditional models use equations; agent-based models simulate behavior through computer rules.”
Some key findings of the study:
- In households in areas with higher population density, odds someone from the household would bring COVID home and spread it within the household were higher.
- In lower-density areas, the virus wasn’t as likely to enter the home, but once it did, it was even more likely to spread.
- Regardless of the setting, human behavior — like compliance with mask mandates and quarantines — had the biggest impact on transmission patterns.
- The dynamics of infection waves differed according to population density.
Agusto said her investigation originated in collaborations with scientists at the Auburn University-based SAMSA-Masamu Program, which aims to enhance research ties between the U.S and Southern Africa in mathematical sciences and related areas.
“At Auburn University, Overtoun Jenda and his colleagues, including Suzanne Lenhart — who is a co-author on the paper with me — had a vision to create a collaboration between Africans and Americans. They would take Americans to Africa to collaborate with them on different topics,” Agusto said. “I was part of that group during COVID, working with Suzanne, who was my postdoc adviser. Well, we did not go physically. We were in ‘Zoom land’ during COVID, working virtually. Subsequently, I have gone with them as a U.S. scientist — even though I am originally Nigerian — to Mozambique and then South Africa.”
Agusto’s co-authors were Inger Fabris-Rotelli of the University of Pretoria in South Africa, Christina Edholm of Scripps College in California, Innocent Maposa of the University of the Witwatersrand in Johannesburg, Faraimunashe Chirove of the University of Johannesburg, Chidozie Chukwu of Georgia Southern University, David Goldsman of the Georgia Institute of Technology and Suzanne Lenhart of the University of Tennessee.
Agusto’s research was supported by a National Science Foundation Incorporating Human Behavior in Epidemiological Models (IHBEM) grant, for which she serves as principal investigator, involving four US institutions: KU, George Mason University, the University of North Carolina at Greensboro and Inter American University of Puerto Rico.
Journal
PLOS One