Twitter Analytics Project HealthMap Outperforming WHO in Ebola Tracking


HealthMap, a collaborative data analytics project launched in 2006 between Harvard Medical School and Boston Children’s Hospital, has been quietly tracking the recent Ebola outbreak in Western Africa with notable accuracy, beating the World Health Organization’s own tracking efforts by two weeks in some instances.

HealthMap aggregates information from a variety of online sources to plot real-time disease outbreaks. Currently, the platform analyzes data from the World Health Organization, Google News, and GeoSentinel, a global disease tracking platform that tracks major geography changes in diseases carried through travelers, foreign visitors, and immigrants. The analytics project also got a new source of feeder-data this February when Twitter announced that the HealthMap project had been selected as a Twitter Data Grant recipient, which gives the 45 epidemiologists working on the project access to the “fire hose” of unfiltered data generated from Twitter’s 500 million daily tweets.

By combining these sources of information, HealthMap is able to track real-time infectious disease outbreaks. The map plots outbreaks in a way similar to Google Flu Trends, but goes far beyond the narrow scope of Google, which only tracks Flu and Dengue Fever. During the recent Ebola outbreak, HealthMap was one of the first to pick up on a problem in the area. On March 14, HealthMap was tracking the development of a “mystery hemorrhagic fever” in Guinea, five days before Guinea’s Ministry of Health confirmed an unidentified outbreak in the region, and nine days before the WHO issued its first warning on the Ebola outbreak. The platform has since tracked new individual cases and Ebola-related deaths as the outbreak spread across Guinea, and into Sierra Leone, Liberia, Nigeria, and beyond. By March 24, HealthMap had identified 59 Ebola-related deaths across the region. The platform is now attributing nearly 1,000 deaths to the disease.

While real-time surveillance tools like HealthMap are a novel and interesting use of otherwise arbitrary data sources, like Twitter feeds, the real work in syndromic surveillance still starts and ends with health workers on the ground in the affected area. The current Ebola outbreak is now thought to have started months ago, in a small Guinea village called Guéckédou. There, a small boy, thought to be patient zero, died from Ebola-like symptoms in December 2013. Shortly after, his mother passed away, and then his grandmother grew ill and died. There was a large funeral for the family that drew mourners from Guéckédou as well as from neighboring towns and villages. After the funeral, when attendee’s returned to their villages, they brought Ebola with them, and the outbreak was born.

Cases like these, so far from an active population of Twitter users or sick patient’s Googling tell-tale key words, remain invisible to tools like HealthMap until the outbreak has established itself and local health workers monitoring the region begin to put the pieces together. Then, the tweets and Google searches start to increase to the level that tools like HealthMap can see and track the problem. If ever there was a use case for a cheap symptom reporting tool for rural communities like these, this would be it. If local health workers had known sooner that three members of a family had died from an Ebola-like sickness, containment measures may have followed faster. Epidemiologists may not ever be able to stop the initial exposure, but if they can respond in time to stop a funeral, they might be able to prevent the outbreak.

Enjoy HIStalk Connect? Sign up for update alerts, or follow us at @HIStalkConnect.

↑ Back to top

Founding Sponsors

Platinum Sponsors