When facing a pandemic or other health threat, public health decision makers have a host of what-ifs to consider. What if we run out of hospital beds? What if we close schools or workplaces? What if the virus mutates?
Dr. Bruce Lee, assistant professor of medicine, epidemiology, and biomedical informatics at the University of Pittsburgh, provided an unusual service during the last H1N1 outbreak. Embedded at the US Department of Health and Human Services, Lee was on call to model situations that had decision makers concerned. Using a high-powered computing platform that he compares to game worlds like SimCity, he was able to provide visualizations of different scenarios so that decision makers could better weigh their responses to changing information.
“People can get lost in numbers. But if you show them something with a striking visualization, they perk up. It gets their attention,” says Lee. Modeling also improves communication among specialists who may have different areas of expertise. When medical experts and policymakers have to reach decisions based on best available information, “a picture really is worth a thousand words.”
Using an approach called agent-based computational modeling, Lee and his colleagues work with a virtual population lab that can mimic actual population patterns in a specific part of the world or even globally. Using US census data, for instance, they can run simulations to show how different health scenarios would affect the nation’s 100 million households.
One project underscored the importance of treating low-income populations first to achieve benefits across society. Poor people tend to use public transportation and travel greater distances to work than more well-heeled populations, making them more likely transmitters. “Moral and ethical arguments are effective when you want to work for some type of social change,” Lee says, “but sometimes it’s even more effective to have a utilitarian argument.”
Lee, who has an MBA and an MD, leaves it to others to suggest which questions to analyze. “My interest is using modeling to address real-world questions. We work closely with decision makers to understand, what are the questions they need answered? What are the challenges to answering those questions?” He is also part of a network of modelers known as MIDAS, for Models of Infectious Disease Agent Study, started by the National Institute of General Medical Sciences (NIGMS) in response to 9/11 and associated health threats.
Agent-based modeling is an increasingly useful tool to analyze a range of public health issues, including bioterrorism threats. “The limitation is how quickly people can be trained to use these tools,” says Dr. James Anderson, program director of the Division of Biomedical Technology, Bioinformatics, and Computational Biology at NIGMS. Anderson says the greatest value for modeling comes at the planning stages, “when you can think about something that hasn’t happened yet.”
Still relatively new, agent-based modeling requires both powerful computers and specialized understanding. Lee credits his modeling skills to an interest in gaming that dates to his youth, followed by business school classes in modeling and business analysis, and then a stint at Quintiles, a clinical research organization, where he did economic modeling for big pharmaceutical companies.
Lee and his colleagues are currently studying vaccine supply chains in a project for the Bill & Melinda Gates Foundation, and also collaborating with UNICEF and the World Health Organization. “People who need vaccines the most are sometimes the last to get them,” Lee says, “especially if they’re in remote areas or underserved populations. Models can help inform decision making about the best way to get vaccines to people.”