The global incidence of dengue is estimated at 390 million cases per year worldwide . Endemic in many Asian and Latin American countries, dengue has become a leading cause of hospitalization and death among children in these regions  and contributes to substantial economic loss for governments and households . Real-time dengue surveillance, therefore, is critical for identifying areas where transmission is ongoing or likely to occur so that interventions can be optimized.
Despite the health and economic impacts of dengue, population-level control methods are limited, resource intensive, and largely ineffective to date. Meanwhile, new data sets from Internet users’ activities have been shown to capture social mechanisms that frequently help fill epidemiological surveillance information gaps.
These data sets include social media monitoring (Twitter microblogs or Facebook posts), tracking of Google search patterns, monitoring mobile phone use, and crowd-sourced participatory disease surveillance tools.
The HealthMap team at Boston Children’s Hospital has been successful at developing accurate methods to produce real-time and forecast estimates of flu incidence using Google search data [4,5,6], Twitter , Wikipedia , crowdsourced participatory disease surveillance tools (such as Flu Near You) , and clinician’s databases (such as UpToDate) .
These methods produce real-time estimates of flu activity that closely track the flu incidence as reported by traditional public health surveillance systems, such as Influenzanet (CDC) in multiple location within the US. The HealthMap team has explored some of these approaches to track Dengue [11,12] in other geographic locations, and continues to improve upon these methodologies.
Building on these opportunities for innovative fight against dengue, HealthMap and Break Dengue are partnering to develop a complementary and innovative surveillance platform using digital resources to generate new data and disseminate it down to the populations at risk. The goal is to continue incorporating additional disparate data sources to improve our dengue incidence estimates.
Our vision is to launch an Internet platform that would display real-time estimates (and forecasted ones) of Dengue activity in multiple locations around the world by leveraging disparate data sources including: traditional clinical reporting systems, crowd sourced disease surveillance tools, internet-based services such as Google, Twitter and Wikipedia.
In addition, our aim would be to empower local authorities with actionable information that may help them establish targeted alert systems to prevent further disease spread. As a pilot effort, we are applying an improved method based on Google searches to predict dengue activity in two pilot locations: Brazil and an Asian country (see Yang et al. 2015). We anticipate that our new methodology will produce more accurate estimates than those previously available through (the now-discontinued) Google Dengue Trends tool.
Call for partners
Break Dengue and HealthMap are looking for partners to support and participate in the optimization of the dengue surveillance tool in the implementation of the next steps of the program, namely (a) displaying dengue predictions on the HealthMap website and (b) identifying and integrating additional data sources to track the burden of dengue in real time. These data sources may include 16 Twitter microblogs, and/or local hospital data.
Integration of a mechanism to display likely mosquito population locations and migration patterns could also be envisaged in the long term.
Please find my presentation at ISNTD Bites last month in London here:Attached files: 20160316_BreakDengue_ISNTD.pdf
and here is an updated visual about the data flow we have in mind:Attached files: dengue_multi_and_crowd-sourced_participatory_disease_surveillance_tools_3Xp0Rpg.jpg