Influenza A (H7N9) and the Importance of Digital Epidemiology

Salathé M, Freifeld CC, Mekaru SR, Tomasulo AF, Brownstein JS. Influenza A (H7N9) and the Importance of Digital Epidemiology. New England Journal of Medicine. 2013 Aug 1;369(5):401–4.

When Hampton et al. outline their vision for “open ecology,” their manifesto includes accessibility of large data streams, to prioritize data stewardship over data ownership. These data can indeed come from large, collaborative projects such as NEON, but “open ecology” might also utilize more passive sources of information. Even if we roll our eyes at how “tweet” is included in standardized dictionaries, we can’t deny that social media is powerful; furthermore, we might be able to use  digital data ranging from published article alerts to buzzworthy news events to gain some scientific insights about the way information spreads through populations or how networks respond to perturbations. In this perspective piece, Marcel Salathé and colleagues discuss how social media is informing a subfield of infectious disease dynamics, digital epidemiology, and how these open data streams can aid in the detection of outbreaks, the monitoring of infections, and the assessment of behavioral responses to epidemics.

The authors here focus their discussion on the public (health) response to an influenza A outbreak in China, in which 132 persons have been infected with H7N9 since 2013.  The outbreak hasn’t yet reached terrifying proportions , but the virus itself is rather novel, difficult to detect in natural avian hosts, and genome sequencing  suggests the potential for human-to-human spread. For these reasons, the public health response in China has been rather vigilant, dedicating substantial resources to traditional outbreak response: running the laboratory tests, tracing the contact network, and so on and so forth. Salathé here argues that utilizing digital sources of information can greatly complement these standard methods, specifically because data gleamed from Twitter, HealthMap, blogs, and email listservs  comes from personal information. And personal information can be good information when it allows us access to firsthand, rapid response data. As an example, a hospital employee first uploaded an image of the medical record of a patient with H7N9 infection to Twitter, alerting the Chinese government to the initial case faster than through traditional medical avenues. Similar sharing of information via social media networks assisted real-time monitoring of the outbreak and epidemic dynamics. Relatedly, a time-series analysis of these data streams helped epidemiologists to determine that the initial H7N9 case above was indeed the first report of the novel infection, given that social media activity about influenza-related symptoms in China was quite low beforehand. The authors also note that matching the digital information time series with the hospital record time series works quite well, in that both follow general epidemic curve patterns (although interestingly, the spike in Tweets precedes the spike in cases by a week or so).

Digital epidemiology certainly has its fallbacks (e.g., interpreting metaphors in tweets), but the authors make a nice case for how use of open and real-time data streams can enhance transparency and help health officials in understanding outbreak dynamics. The general concept of using social media in our analyses of social trends or disease emergence must have taken some foothold, as there are some solid R packages (twitteR ) and decent tutorials on the matter. And while these data streams do rely on the internet and thus a certain access to resources, complementing digital sources with more general data from rural zones should substantially improve our understanding of disease emergence, spread, and management (e.g., mobile phone data informing malaria control).

About The Author

Add a Comment