Table 1.
Number of official and non-official events of AI detected by PADI-web and HealthMap between July 2018 and June 2019.
The number of non-overlapping events is shown between parentheses.
Fig 1.
Timeliness in the detection of AI outbreaks according to the type of host for A) PADI-web and B) HealthMap. For visibility, extreme values i.e., less than 30 days and higher than 30 days are not shown.
Table 2.
Types of sources (i.e., nodes) in PADI-web and HealthMap networks disseminating outbreak-related news on Avian influenza between 1 July 2018 and 30 June 2019.
Fig 2.
Performance of sources in terms of A) in-degree and B) out-degree, aggregated by type. The y-axis has been log-scaled. Distributions of in-degree and out-degree are represented with box plots based on a 95% confidence interval (outliers are represented with dots).
Table 3.
Top-5 sources in terms of in-degree, out-degree and all-degree.
The EBS tools PADI-web and HealthMap were excluded as they were chosen as the aggregators in our study.
Fig 3.
Proportion of the types of primary and secondary sources according to their role in the (a) PADI-web and (b) HealthMap late detection networks. Primary sources are sources that are the first to emit an event, secondary sources are sources which receive and emit an event to another source.
Fig 4.
Proportion of the types of primary and secondary sources according to their role in the (a) PADI-web and (b) HealthMap early detection network. Primary sources are sources that are the first to emit an event, secondary sources are sources which receive and emit an event to another source.
Fig 5.
Proportion of the geographic scope of primary and secondary sources in the PADI-web and HealthMap early and late detection networks.