As reported in Wired, machine learning programs are being used by researchers to scan social media posts, news reports and websites as an indication of potential symptoms.
Artificial intelligence researchers are applying machine learning techniques to social media and other data for signs of where the disease may be spreading. The coronavirus which became prominent in Wuhan, China at the end of 2019 has triggered a global health emergency. Experts are still unsure of how contagious the virus is but infections and deaths continue to increase. More than 31,000 people have contracted the virus in China alone and 630 people have died worldwide.
The machine learning program is looking for social media posts that make reference to specific symptoms known to be displayed when contracting the virus such as respiratory problems and fever from geographical areas where doctors have reported potential cases. Intelligent processing is used to distinguish between a person discussing news of the virus and a person discussing how they feel. John Bownstein, chief innovation officer at Harvard Medical School is an expert on mining social media for health trends and is part of the team using the machine learning program. Bownstein made reference to how critical it is to determine where the virus may surface to enable the authorities to allocate resources in the right places and control the spread.
Brownstein stated the technique could help experts learn how the virus behaves. There is potential to determine the age, gender and location of those most at risk more quickly than other sources. Crowdsourced information gathered by volunteers or via websites set up to provide information about the virus is also extremely important. Brownstein is working with a Boston-based company, Buoy, that offers health advice to millions in the US online and through portals. Buoy will provide advice to those suspecting they may have the virus whilst feeding it to Brownstein and others as another data source.
Andy Tatem, a professor at the University of Southampton has used anonymized historical data from smartphones to model how the virus may have moved out of Wuhan in the days after it appeared. It may be possible to track individuals using the movement of their smartphones, Tatem believes this is less useful than understanding broader trends and dynamics. Its unclear how widely the virus might travel, but Tatem says the biggest concern is that it could appear in countries with fewer healthcare resources to effectively take action.