WVU researchers deploy AI to target COVID testing in low-vaccination areas

Testing was supposed to be the backbone of pandemic response. The National Institutes of Health had called it "the key to getting back to normal." But across the country, testing rates fell sharply even as new variants spread. In late 2020 and early 2021, when vaccines first became available, testing dropped by a third—a collapse that rippled through everything epidemiologists needed to do: track the virus's true spread, calculate how fast it moved from person to person, plan where resources should go.

At West Virginia University, two researchers saw an opportunity in that gap. Brian Hendricks, an assistant professor of epidemiology and biostatistics, and Brad Price, who teaches machine learning at the business school, began asking a different question: What if you could use artificial intelligence to figure out exactly where testing mattered most? The National Institute on Minority Health and Health Disparities awarded them $2.15 million to find out.

The logic was straightforward but powerful. Vaccination rates vary wildly across West Virginia. Some counties have high uptake; others have far less. In those low-vaccination areas, the virus finds more fertile ground. Outbreaks simmer locally. But without testing, no one knows they're there. Hendricks explained the stakes plainly: "A drop in testing hurts your epidemic modeling, your calculation of the basic reproductive number, your ability to plan for research allocation. As the pandemic evolves, we have to keep testing to monitor localized outbreaks and understand when a new variant is introduced."

Machine learning offered a way to cut through the noise. Unlike static analysis—a snapshot of data frozen in time—machine learning systems absorb constantly updating information and grow more accurate as they learn. Price emphasized why this mattered for a moving target like COVID: "We've seen variants pop up. We've seen surges in cases. We've seen cases fall off. We've seen masks go on and come off. And now we're talking about booster shots. If we're just saying, 'This is the data. Analyze it,' without considering how it's moved over time and how it will continue to move over time, we're missing a big piece of the puzzle." The researchers would build tools that tracked case counts, testing trends, emerging variants, and vaccination data together, then predict which counties faced the highest outbreak risk.

But prediction alone wasn't enough. Hendricks and Price knew that getting people to actually show up for tests required understanding why they weren't showing up in the first place. They began conducting interviews in low-vaccination communities, asking directly: What barriers kept people from testing? How did residents feel about it? What might motivate them? The answers would shape the intervention itself.

Their approach rejected the idea that one strategy fits everywhere. Instead, they imagined something more human-scaled. Picture a free community cookout, advertised on social media, with local first responders circulating among the crowd. Staff from QLabs, a research partner, would be there to conduct tests. Hendricks described the vision: "I want them to do what they do every day, which is go up to the people who are eating the food at these events and say, 'Hey, I care about you. How's your family doing? How's your mom doing? Have you gotten tested lately? You haven't? Well, I care about you. Let me walk you up to the table where you can get tested.'" The point was trust, not mandate. A familiar face asking a familiar question.

Sally Hodder, who oversees clinical research at WVU, underscored why this mattered now. Unvaccinated people faced far higher risk of severe disease and death. But new oral antivirals had changed the equation—if someone tested positive, they could take pills that sharply reduced hospitalization risk. Testing, in other words, had become a gateway to treatment, not just surveillance. The work was urgent.

Price looked further ahead. This wasn't the first pandemic, and climate change was likely to bring more infectious diseases in the years to come. The infrastructure they were building—the data systems, the protocols, the ability to move quickly—could matter for whatever came next. "At the beginning of the pandemic, we couldn't do anything because we didn't have data," he said. "In the middle of the pandemic, we couldn't do anything because we didn't have an infrastructure for that data. Now we're starting to piece it together. I'm going to be focusing on making sure we have that infrastructure so that the next time this happens, we have our policies, protocols and systems built, and the second we have data available, we can hit the ground running."