An artificial intelligence (AI) model that relies on judgment variables accurately predicts COVID vaccine uptake, which could be a useful and powerful tool for public health officials in future vaccine campaigns. Researchers from the University of Cincinnati and Northwestern University detailed their findings this week in the Journal of Medical Internet Research (JMIR) Public Health and Surveillance.
Study compared demographic, judgment variables
First, they collected demographic information from 3,476 adults who responded to an online survey in 2021, when the COVID vaccine had been available for about a year. The group's profile was similar to the general US population, based on Census Bureau figures. About 73% of the respondents said they were vaccinated, slightly higher than the 70% for the general population.
The researchers also asked if participants followed other COVID recommendations, such as masking, social distancing, and hand hygiene.
Then they asked participants to perform tasks that measured people's judgments, measuring factors such as risk aversion and aversion to loss. Participants were asked to rate how well they liked or disliked a set of 48 emotionally evocative pictures, ranging from sports to cute animals. A mathematical model designed to identify reward and aversion judgments quantified each peson's responses.
The team compared the demographic and judgment variables between the vaccinated and unvaccinated groups, then assessed how well three different machine-learning approaches predicted whether participants got the vaccine. They found that 7 demographic and 15 judgment variables predicted vaccine uptake.
Simple tool for a complex public health challenge
The researchers said advantages of the picture-rating task are that it relies on a small set of data that can be collected, processed, and stored in databases ahead of time, and that data collection can’t be biased against vaccines because the task has no perceivable tie-in to vaccine choices. They noted that public health officials could used the data to predict vaccine uptake locally or nationally, which would help with vaccine rollouts, planning supply, and targeting messaging to areas where uptake will likely be low.
An AI-guided method can make accurate predictions without expensive and time-consuming clinical assessments, the team wrote.
In a University of Cincinnati press release, Aggelos Katsaggelos, PhD, one of the paper's senior authors and endowed professor of electrical engineering and computer science at Northwestern University, said the study goes against the big-data grain. "It can work very simply. It doesn't need super-computation, it's inexpensive, and can be applied with anyone who has a smartphone," he said.
