An assortment of mathematicians, modelers and sociologists at University of Idaho have been working since the spring to predict the spread of COVID-19 in rural communities. The Idaho Pandemic Modeling Group, part of UI’s Institute for Modeling Collaboration and Innovation, makes models based on local data. The group has been working for supplemental funding attached to IMCI’s current National Institutes of Health grant.
“We are a collection of people and we’re working on different aspects of the problem,” Craig Miller, IMCI’s modeling core director, said. “If we get funded, our plan is to develop models that are more tailored to individual communities that describe individual communities more accurately and try to understand the likely spread of the virus in the individual communities.”
According to the group’s website, the team uses two models. One of these is a “Susceptible, Exposed, Infectious and Recover” or SEIR model. SEIR models demonstrate how contagions spread in a population.
The group uses a metapopulation SEIR model based out of 50 Idaho cities in five age categories to. In this type of model, different behaviors and the effects of social networks are often evened out.
The agent-based model the group uses is significantly more detailed than the SEIR model. It can show the probability of interaction and infection, which means the group can assess intervention strategies. This captures information about behaviors and social networks that the SEIR model averages out.
“The reason an agent-based model is important (in rural communities) is because when the populations are really small, it’s hard to capture the dynamics with (a SEIR model),” Jennifer Johnson-Leung, an associate professor of mathematics at UI, said.
IMCI’s modeling website has various graphs and tables with the status of COVID-19 on UI’s campus, estimated 10-day infection totals by county across Idaho and similar graphs for communities near the state border.
“What’s complicated about modeling is not so much the mathematics,” Johnson-Leung said. “It’s the science of it.”
Johnson-Leung said the mathematical theory used for their models is not suited for the large changes in human behavior seen throughout the pandemic, which creates a small gap between actual cases and what the data shows in the group’s models. The team is looking for ways to fill the gaps caused by behavior changes, including analyzing a survey sent to people in Idaho, Texas and Vermont to get an idea of risk perception and behavior choices in different communities. The team is starting to receive the first results from the survey, Johnson-Leung said.
“In the scope of behavioral choices some people, of course, don’t have a lot of choices in their behavior because they are hemmed in by economic constraints,” Miller said. “They have to go to work, they have child care issues, all those things that lots of people deal with.”
Miller said the survey would be looking for things like how seriously people take safety protocols or quarantine and whether they would seek out a COVID-19 test after coming in contact with it or showing symptoms.
Idaho Governor Brad Little rolled Idaho back into Stage 2 of Idaho Rebounds on Friday, making the types of choices in the survey even more relevant to the state’s residents.
“We have a fragile network of rural health clinics serving a high risk older population,” Johnson-Leung said. “I’m afraid that we’re going to see a lot of death in our rural areas.”
Idaho’s hospitals are being overwhelmed to the point of not being able to answer phone calls because of the scope of the spread of COVID-19, according to the Associated Press.
After working with COVID-19 statistics since last spring, Johnson-Leung said the weight of what she was looking at hit her suddenly while sorting data last week.
“(The weight of) all of these people that have gone to the hospital and so much fear and suffering, it’s frustrating,” Johnson-Leung said. “I feel like we, in this region, weren’t surprised by this. We knew this was coming, this is not shocking. What’s happening right now was absolutely predictable, and was predicted. It’s heartbreaking.”
Anteia McCollum can be reached at [email protected] or on Twitter @antxiam5.