September 17, 2024

Initiative-funded AI pilot project looks to create datasets that spot and predict future pandemics

The origins of the COVID-19 pandemic reflect a larger trend of increasingly frequent, severe and widespread human disease transmission caused by changes in global biotic and abiotic systems. These changes are a result of climate change, land-use change, globalization and urbanization, all of which are causing unprecedented shifts in ecosystems that result in a greater risk for zoonotic spillover.

In summer 2024, a team of interdisciplinary researchers at the University of Washington were awarded an artificial intelligence-focused pilot grant from the Population Health Initiative to create new, dynamic datasets that accurately identify key risk factors for pandemic emergence in Washington State. These datasets are in large part an effort to restore the precision of models that recognize “hotspots,” or areas of high risk for disease spread, and expand the potential for these models to accurately predict future “hotspots.”

“About 15 years ago, there was a lot of interest in creating hotspot maps of zoonotic spillover using variables like precipitation, temperature, human population density, and wildlife biodiversity, and linking those to places where spillover has happened with a machine learning model,” shared Julianne Meisner, lead co-investigator and an assistant professor of Global Health. “The challenge is that these maps often identify many thousands of hotspots because they have to simplify things a lot, which indicates that these variables are not suited for the task of identifying zoonotic emergence.”

“This project is trying to rectify these models by starting off with maps that are much more suitable with datasets on landscapes and human movement that are much closer linked to what we know causes zoonotic emergence so we can do a better job of modeling and have better input data,” said Meisner.

The project’s team includes researchers from Paul G. Allen School of Computer Science & Engineering, the School of Public Health and the Department of Industrial & Systems Engineering.

As part of the project’s aim to create high-resolution datasets, Meisner and her research team will utilize key landscape characteristics like biodiversity, animal and human movement and land-use to better understand and forecast disease spread as the result of global phenomena.

“People have been living in close proximity to wildlife and forests for a millennia and I think there is an increase in risk of disease spread that is happening because of forces that are global,” explained Meisner. “For example, climate change could be leading to ecosystem stress that is increasing the risk of an activity that people have always done, but now it’s riskier than it used to be. There’s also land-use change that involves deforestation, wildfires, conversion of land biomes like savannah to grasslands, which are being caused by globalization or urbanization that ultimately strains ecosystems and causes diseases to emerge.”

The project’s measures of success rely on the successful presentation of their datasets to stakeholders in partnership with the Washington State One Health Collaborative, as well as the establishment of multidisciplinary mentoring meetings with the project’s PhD student.

The project began on August 1, 2024, and will last for one year, with the goal of producing proof-of-concept work that will allow the team to secure additional funding from the Advanced Research Projects Agency for Health (ARPA-H), the National Science Foundation (NSF) or the National Institutes of Health (NIH).

“We’re starting it in Washington state but would like to expand it to be a global effort. We’re trying to work out the kinks and figure out how this works, if we can do this effectively and if we can produce more valid maps,” said Meisner. “We also have to come up with some creative ways to get around the fact that zoonotic emergence is quite rare, but our goal is to package this in a larger proposal to a funder such as ARPA-H, NSF or NIH in order to keep this going and be a multidisciplinary training opportunity.”