Crowdsourcing Platforms Monitor Disease, Dissent, and Disasters

Epidemiology pioneered crowdsourcing with data from "informal" sources such as Internet keyword searches in the first decade of the 21st century, gaining insights before data from traditional sources such as sentinel providers could be compiled. Now, a veteran of those online biosurveillance efforts has applied his expertise with crowdsourcing to catalog the effects of political strife, combining platforms developed for epidemiological compilation and crisis reporting to create Humanitarian Tracker.

The effort’s founder, Taha Kass-Hout, M.D., has been developing public disease surveillance platforms, including eQuest, a Web-based survey creation-and-analysis tool for epidemiologic and disease outbreak investigation, and Riff, an open source social networking platform for integrated early warning and response, for better than a decade.

Dr. Kass-Hout says the unaffiliated and non-partisan Humanitarian Tracker is a logical extension of Syria Tracker, a crowdsourcing effort focused on the civil war in Syria. Syria Tracker was founded in 2011. "After almost two years of documenting the events live from the ground using crowdsourcing and other media tools," Dr. Kass-Hout says, "it was decided that others around the world could be helped as well using a similar platform, and hence, Humanitarian Tracker was established."

The Humanitarian Tracker team, all volunteers, uses a hybrid platform of open crowdsourcing technologies that combine eyewitness "citizen journalist" entries as well as reports from traditional news organizations, including:

• Ushahidi, which collects citizen data on its Crowdmap platform from multiple channels including email, Twitter, YouTube videos, online news, syndicated feeds like Geocoded Really Simple Syndication, Webform, or mobile apps. Dr. Kass-Hout says the Ushahidi team has been "very accommodating to address issues that evolved over time, such as security, cloud hosting, and adding new features such as download, or subscribe to alerts based on location or specific tag of interest." Ushahidi was initially developed to map reports of violence in Kenya in early 2008 following a disputed presidential election.
• A modified version of the HealthMap disease surveillance platform. The HealthMap technology, developed at Boston Children’s Hospital in 2006, automatically monitors global disease reports using both official public health data and commercial news feeds, Google Maps, and Google translation. Dr. Kass-Hout says Humanitarian Tracker’s HealthMap platform uses a modified version of the platform’s disease detection algorithm to accommodate different key searches, particularly killings due to armed violence and rape. HealthMap provides approximately 10 Google News-generated reports of violence per day. Sources sympathetic to the Syrian government, which provide information on killings carried out by opposition groups, are also included.
• CrisisTracker, which collects and organizes reports from Twitter. "Twitter is an interesting complement to traditional news media and eyewitness reports, as news spreads extremely quickly and at least in theory, the report coverage is good enough to leverage Twitter’s user base as a form of distributed sensor network," says Jakob Rogstadius, CrisisTracker’s originator. The system automatically tracks a set of keywords on Twitter and clusters tweets based on their word similarity. Information about who and how many people shared a piece of information can be used to estimate how important the information is, and to whom, without requiring a computer to first understand what the information is about.

"There is no silver bullet solution, so you have to focus on each system or tool’s strength and study carefully how and when to put them together," Dr. Kass-Hout says. "Ushahidi’s effectiveness depends entirely on the size, coordination, and motivation of crowds which adapts well to needs of specific disasters, but is difficult to scale to match information inflow rates during very large events like what is happening in Syria. HealthMap’s strength has been its ability to mine more contextualized reports or news in English, but it needs further refinements to help with less contextualized data, like Twitter, and support for Arabic language."

"The Syria Tracker people have provided very insightful and valuable feedback regarding how well CrisisTracker supports remote monitoring of complex large-scale events," Rogstadius says. "Much of their feedback has been written up in an upcoming paper to be published in the IBM Journal of Research and Development, and an updated version of CrisisTracker is currently in development with expected release around June [2013]."

Dr. Kass-Hout says the team also uses multiple methods to vet reports. Moderators review submissions, either before or after public dissemination, and users can provide feedback and corroboration of submissions with other sources.

"Social media is, by nature, a venue for two-way information exchange where crowds can verify and evaluate the quality of information shared by other users when appropriate," he says. "The team works hard to weed out duplicates by comparing the images, video, and written statements associated with each event."

True to its founders’ epidemiological roots, Humanitarian Tracker also includes a discrete epidemic platform, Epidemic Tracker, which tracks cases of tuberculosis, Leishmaniasis, and viral hepatitis.

"Disease and water and food tampering were among the first categories we had in Syria Tracker," Dr. Kass-Hout says. "Epidemic Tracker was emerged to focus just on diseases."

Data from Syria Tracker is already being used in official information sources, such as maps from the United States Agency for International Development (USAID), though one veteran of crisis crowdsourcing says official hesitance to rely on such "alternative" data remains the norm — just as public health officials have been hesitant to rely on crowdsourced data to announce epidemics.

"In general, the decision-making structures of humanitarian organizations are not geared towards leveraging alternative and unorthodox sources of information," says Patrick Meier, former co-director of Harvard Humanitarian Initiative’s Crisis Mapping and Early Warning program, and former Director of Crisis Mapping at Ushahidi. "Compounding this problem is the fact that there is little evaluation of traditional decision-making processes. So while the data may be actionable and credible, humanitarian organizations are still struggling to turn this data into action."

Meier does see one scenario in which government agencies and NGOs are likely to adopt such crowdsourced data more quickly.

"In contexts similar to Syria — and Libya during early days of the crisis there — where access is highly restricted, yes, I do see agencies turning to these unorthodox information sources to inform their situational awareness."