How AI could help stop the next pandemic before it starts

The artificial intelligence tool is the first to use large language modeling to predict infectious disease risk

Could artificial intelligence tools be used to stop the next pandemic before it starts?

During the Covid pandemic, new technology developed by researchers at Johns Hopkins and Duke universities didn’t exist. But, for the first time, researchers there say they’ve devised a revolutionary large language modeling tool - the type of generative AI used in ChatGP - to help predict the spread of any infectious disease, such as bird flu, monkeypox, and RSV. That could help save lives and reduce infections.

“Covid-19 elucidated the challenge of predicting disease spread due to the interplay of complex factors that were constantly changing,” Johns Hopkins’ Lauren Gardner, a modeling expert who created the Covid dashboard that was relied upon by people worldwide during the pandemic, said in a statement.

“When conditions were stable the models were fine. However, when new variants emerged or policies changed, we were terrible at predicting the outcomes because we didn’t have the modeling capabilities to include critical types of information,” she added. “The new tool fills this gap.”

Gardner was one of the authors of the study published Thursday in the Nature Computational Science journal.

Researchers say they’ve developed a revolutionary new artificial intelligence tool to help predict the spread of infectious disease. Could it predict the next pandemic before it starts? (Getty Images)

The tool, named PandemicLLM, considers recent infection spikes, new variants, and stringent protective measures.

The researchers utilized data that had never been used before in pandemic prediction tools, finding that PandemicLLM could accurately predict disease patterns and hospitalization trends one to three weeks out.

The data included rates of cases hospitalizations and vaccines, types of government policies, characteristics of disease variants and their prevalence, and state-level demographics. The model incorporates these elements to predict how they will come together and affect how disease behaves.

They retroactively applied PandemicLLM to the Covid pandemic, looking at each state over the course of 19 months. The authors said the tool was particularly successful when the outbreak was in flux. It also outperformed existing state-of-the-art forecasting methods, including the highest performing ones on the Centers for Disease Control and Prevention’s CovidHub.

“Traditionally we use the past to predict the future,” author Hao “Frank” Yang, a Johns Hopkins assistant professor of civil and systems engineering, said. “But that doesn’t give the model sufficient information to understand and predict what’s happening. Instead, this framework uses new types of real-time information.”

There will be another pandemic. The researchers hope this can support public health response (Getty Images)

Going forward, they are looking at how large language models can replicate the ways individuals make decisions about their health. They hope that such a model would help officials to design safer and more effective policies.

More than a million Americans have died from Covid. It’s not a matter of if there will be a next pandemic, but when. Right now, the U.S. is dealing with the spread of H5N1 bird flu, RSV, HMPV, pertussis, and measles, among other health concerns. Vaccination rates for measles have plunged since the pandemic, and general vaccine hesitancy has increased. That has resulted in fears that the nation could see decades of health progress reversed. Furthermore, U.S. health officials have acted to separate from global efforts to respond to pandemics, withdrawing from the World Health Organization earlier this year. Last month, they limited access to Covid vaccines for certain groups.

“We know from Covid-19 that we need better tools so that we can inform more effective policies,” Gardner said. “There will be another pandemic, and these types of frameworks will be crucial for supporting public health response.”