Biology Reference
In-Depth Information
As shown in Figure 6.1, the HealthMap system gathers alerts, classi-
fies them by location and disease, stores them into a database, and then
displays them to the user. As raw data are loaded from the Web, it is con-
verted into a standard “alert” format. This standard format consists of a
headline, alert issue date, brief summary of the alert, and text that is fed
into a parsing engine for an initial classification. The classification engine
determines the principal diseases and locations associated with an alert.
Upon being classified by disease and location, the alerts are stored within
a database designed primarily to support features of the HealthMap Web
application, however, also remain readily accessible for retrospective epide-
miological studies, public health risk mapping, and other research applica-
tions (Freifeld et al. 2008).
More simply put, the HealthMap system characterizes disease outbreak reports
by utilizing a series of text mining algorithms, including the disease and loca-
tion extraction described previously, in addition to algorithms that determine
the relevance of the alert (whether a report refers to a current disease outbreak),
and groups together similar alerts while also removing duplicates (Figure 6.2)
(Brownstein et al. 2008). After categorization by location and disease, articles
are automatically tagged according to their relevance. Tags include (1) break-
ing news (e.g., a newly discovered outbreak), (2) warning (initial concerns of
disease emergence, e.g., in a natural disaster area), (3) old news (reference to a
External Feeds
Loader
Extract Title, URL, Date,
Info Text
Info Text
Ids
Lookup Tree
Parse Location, Disease
DB Inserts
Matching Alerts
AJAX Function
DB
Convert form state to query
Collect results by location
Select Query
Create Marker Calls
Web Page
Display map
Display disease list
Call AJAX function with form state
Form State
Figure 6.1
HealthMap System Architecture.
