Artificial intelligence and machine learning can help improve data-sharing practices for better public health outcomes, a senior government official and an infectious disease epidemiologist said at a GovernmentCIO Media & Research event this week.
Defense Advanced Research Projects Agency's Preventing Emerging Pathogenic Threats (PREEMPT) program, which focuses on preventing diseases in animals and insects from "spilling over" into the human population, provides ripe opportunities for AI and machine learning to identify what factors might affect or cause disease spillover, noted Program Manager and infectious disease epidemiologist Rohit Chitale.
“We have University of California Davis that is actually looking at a disease in West Africa called Lassa fever virus, and how to prevent spillover into humans,” Chitale said at the Artificial Intelligence: New Horizons in Medicine event. “They do that using some really neat AI, machine learning to understand the remote precipitation data, what’s going on in the climate in the area ... and other factors to prevent spillover.”
The two main performers in the PREEMPT program are UC Davis and Montana State University, but Chitale said every one of the six performers are using some form of machine learning to research infectious diseases.
Federal agencies like the Department of Homeland Security can develop practical applications from research conducted by organizations like DARPA, noted DHS CIO Karen Evans.
“If it looks like it's promising to work some of these things, then we get the opportunity to try to implement it within a government enterprise,” she said. “It's taking some of the vision that DARPA has, because you can see the excitement in everything they're talking about, and being able to say is there a practical application now to be able to do that?”
Due to coronavirus, for example, DHS is exploring contact-tracing methods to bring employees back into the office.
“Because of the public health issue we have, there's a lot of opportunities to do that [AI] analysis from that contact tracing that we want to be able to do with our workforce when they come back," Evans said. “[Chitale’s] efforts are going to be powered by a lot of the data sets we already have within the government. There's a lot of promise from this data and this technology, using machine learning and AI to predict trends and those types of things.”
Evans and Chitale also touched on using training sets of data to test new AI algorithms for public health use. Using new AI algorithms on some public health data sets, Chitale said, could raise privacy concerns since the U.S. doesn’t have very well-established privacy standards around government use of personal data.
“You don't necessarily want to start putting out these algorithms and using certain things on machine learning on live data because of all the issues [Chitale] just brought up, dealing with privacy,” Evans said. “Having an environment within an operation like ours, where you can have a data set that you're saying is training data and be able to test, you can see if the machine is actually going to achieve that you want it to achieve.”