Biology Reference
In-Depth Information
4/21 First Report of A(H1N1) in English News Media
4/25 WHO Convenes First Emergency Committee
4/27 WHO Raises Pandemic Alert to Phase 4
4/29 WHO Raises Pandemic Alert to Phase 5
100000
10000
1000
Google insight H1N2
WHO confirmed cases
WHO confirmed deaths
WHO reporting countries
BioCaster news Influenza
100
10
1
Figure 15.1
Log10 scale plots for April 1, 2009, to June 1, 2009, showing (a) Normalized Google Insights
search volume for the term “H1N1” from Google Insight, (b) WHO laboratory confirmed cases
for influenza A (H1N1) (n = 17,410 on June 1, 2009), (c) WHO laboratory confirmed deaths for
A(H1N1) (n = 115 on June 1, 2009), (d) WHO countries reporting A(H1N1) cases (n = 62 on June 1,
2009), (e) cumulative frequency of positively identified news reports found by BioCaster for the
term “Influenza” (n = 4609 on June 1, 2009). WHO data sourced from Epidemic and Pandemic
Alert and Response situation reports for Influenza A (H1N1).
reported human cases of a novel influenza A (H1N1) outbreak centered on
Mexico in seven countries. Subsequently, the WHO formally requested all
countries to activate pandemic preparedness plans and to remain on high
alert for unusual outbreaks of influenza-like illness. The WHO increased its
pandemic threat level for swine influenza to “Phase 5” as reports of multiple
fatalities in Mexico emerged. At the same time, general concern around the
world rose and the information storm in the media moved from reports on
the level of hundreds to the level of thousands. Against this background,
automatic gathering of event-based health intelligence from open media
sources began to take on increased importance as the number of reports
increased rapidly to fill the public's demand for information.
The overall goal of event-based GHISs is to find linguistic signals in unstruc-
tured Web reports for the purposes of near-real-time detection, quantification,
and reporting of public health events to the appropriate authorities. As we show
in this chapter, public health organizations are increasingly considering auto-
matic analysis of natural language data from the Web as well as search engine
data so as to extend traditional indicator-based methods (e.g., notifiable disease
reporting and over-the-counter [OTC] sales monitoring [Ginsberg et al. 2008]).
The rationale for this decision is driven by the need to build global surveillance
capacity. The abundant and near-real-time nature of online news makes it a
cost-effective means of early detection and tracking of health events on a global
scale. Open media sources can bridge the gap between national and interna-
tional surveillance as well as provide timely access to sources on the ground.
