If you’ve attempted to plan a vacation or a wedding recently, you may have looked up how badly COVID-19 is expected to spread.
Hospital administrators and public health officials use these forecasts too, to be prepared for outbreaks.
Most epidemic forecasting models currently used by the Centers for Disease Control and Prevention don’t take into account how people's perceived risks and behaviors such as mask wearing, or social distancing, change over time and how such behavioral reactions influence the spread of the virus. That’s partly because human behavior is tough to predict.
“Things like COVID fatigue is a concept that can influence people’s behavior,” said Navid Ghaffarzadegan, an associate professor at Virginia Tech who specializes in modeling related to human health. He’s on a team of scientists who’ve been studying an interdisciplinary approach to predicting future trends of COVID-19, using data about how people are changing their behavior, as time goes on.
When they compared their models with those used by the CDC, the majority of their predictions were more accurate.
As human behavior changes, these models need to be updated, Ghaffarzadegan says. And even though these types of predictions are more difficult, they could be more useful at revealing what’s ahead. Which could help people better prepare not only for future variants of COVID-19, but for potential other pandemics too.
“And if you don’t incorporate changing behavior in the models, then the accuracy of the predictions won’t be good enough for policy purposes,” Ghaffarzadegan said.
The study “Enhancing long-term forecasting: Learning from COVID-19 models” was published in the journal PLOS computational biology and was co-authored by Hazhir Rahmandad and Ran Xu.