Tracking Disease: Digital Epidemiology Offers New Promise in Predicting Outbreaks

The MSPP, with technical assistance from the U.S. Centers for Disease Control and Prevention (CDC) and its partners, employed many traditional forms of epidemiological surveillance to track the outbreak and try to stem its spread. These included visiting hospitals, interviewing patients in the communities where the disease was first reported, conducting telephone interviews with health facilities, and reporting the results to the press. But they also turned to some nontraditional surveillance methods. Because 86% of the world's population lives under mobile cellular network coverage and mobile phone networks routinely register data that can be used to track the location of all active mobile phone users, researchers were able to track the movement of people in Haiti via their mobile phones, a fact that would become critical in predicting the spread of the disease. Using position data from subscriber identity module cards provided by the largest mobile phone company in Haiti, researchers estimated the magnitude and trends of population movements, allowing for close to real-time monitoring of the disease outbreak. In fact, studies show that Twitter data following the 2010 earthquake accurately tracked the cholera spread, and these data were faster and roughly as correct as the official records detecting the beginning and early progress of the epidemic. Advances in automated data processing and machine learning now allow epidemiologists to meticulously sift through the millions of digital traces we collectively leave behind each day as we conduct our lives online-through Internet searches, social media posts, or the use of our mobile phones. There is a rich treasure trove of data that, though it may provide only a partial glimpse of a disease outbreak, is proving to be quite nimble at capturing the spread of disease. Traditional tracking methods-surveillance of clinics, telephone triage calls, purchases of over-the-counter medications, and school absenteeism, for example-are far more unwieldy. Although researchers caution that these new sources of information cannot completely replace traditional epidemiological methods, they are serving to fill in critical gaps.