Harvard Medical School Professor’s Artificial Intelligence System Maps Coronavirus Outbreak

HealthMap, a Harvard Medical School-developed artificial intelligence technology that can track and predict the spread of infectious diseases, is now being used to monitor coronavirus cases across the globe.

As with previous disease outbreaks, HealthMap — created by Medical School Professor John S. Brownstein and a team of researchers — has helped track reports of the global uptick in coronavirus infections, including in several states across the United States.

Local newspapers and social media posts initially identified early signs of the outbreak before the virus drew international attention, according to Brownstein. The technology collects similar local data and uses it to create an extensive, publicly-available map of the outbreak.

The system updates in real-time as it collects and processes data from the web. Researchers with HealthMap have partnered with organizations including the World Health Organization and Centers for Disease Control and Prevention.

“It provides situational awareness so we get insights globally into what’s happening,” Brownstein said.

Brownstein said the system's goal is not only to identify cases, but also to gather information on the public perception of disease outbreaks.

“The social media mining provides us with a perspective on how the population's reacting to coronavirus and also rumors that people have around the disease,” Brownstein said.

Brownstein founded HealthMap in 2006 with a team of epidemiologists, researchers, and software developers.

Using novel sources of information, including Google searches, social media posts, blogs, and chat room inquiries, the technology identifies markers of various infectious diseases and collects data for researchers and government agencies.

Brownstein said the team created HealthMap to provide clear information on the progression of diseases. He added that it has proven effective in mapping many outbreaks over the past decade, such as swine flu and ebola.

“It was very challenging to get an accurate picture of what was happening around the world in terms of infectious diseases. Either we didn’t know about events, or governments weren’t really reporting them, or it was information that was locked in a whole bunch of different information streams,” he said.

Using a combination of machine learning and crowdsourcing, HealthMap uses information from the web that epidemiologists do not typically use to track diseases, according to Brownstein.

“If you could organize that sort of information — scrape it, tag it, process it, and remove the noise — you could provide a view of emerging outbreaks in a way that hadn't been shown before,” Brownstein said.

Brownstein added that the system can help researchers create complex projections on how diseases may spread.

“When a big event is identified, we spend a lot of time developing, tracking, and building situational awareness, and then feeding that data to make projections, whether that’s through connecting that data to transportation networks or understanding population mobility. So feeding that information to help to form projections of risk,” he said.

—Staff writer Virginia L. Ma can be reached at virginia.ma@thecrimson.com.