Tapping crowds for science: From galaxies to diabetes

Photo: ausnahmezustand/Flickr

What do a project cataloging pictures of galaxies, an RNA folding game, and a call for people with diabetes to contribute data all have in common?

Each is part of a new revolution in science. Called “citizen science,” this revolution takes science out of traditional academic or industrial environments and into the population at large, asking the general public to take part in activities that further particular areas of research.

Citizen science projects tap the aggregate computing power of crowds to help collect or analyze huge data sets, running the gamut from online games (e.g., FoldIt, EteRNA) to screen savers that make use of your computer while it’s asleep (e.g., SETI@home) to projects asking people to count or categorize images from large-scale astronomy projects (e.g., GalaxyZoo, Stardust@home). Some even try to reduce animal-vehicle collisions on the nation’s roadways by cataloging and mapping roadkill.

Citizen science is also starting to extend to medical research. Last year, Harvard Catalyst ran a contest open to every member of Harvard’s extended community – all of the students, faculty, and staff at all of its schools and affiliated hospitals, including Children’s – for new ideas and questions about type 1 diabetes. PatientsLikeMe recently ran a virtual clinical trial on a potential drug for amyotrophic lateral sclerosis (ALS).

Researchers at Children’s are even utilizing crowd power for disease and public health surveillance. Two years ago, a team led by John Brownstein of the Computational Epidemiology Group (CEG) in the Children’s Hospital Informatics Program (CHIP) released an iPhone app called Outbreaks Near Me. In addition to letting users track infectious disease outbreaks on the ground in real time, it enables users to submit an outbreak report. (Outbreaks Near Me integrates with HealthMap, a CEG-developed website that displays a unified and comprehensive view of the current global state of infectious diseases based on data from a range of sources, including on-the-ground reports.) His team has since launched a second app, MedWatcher, that allows users to get drug safety updates and report information about drug side effects.

Most recently, Kenneth Mandl and Elissa Weitzman of CHIP’s Intelligent Health Laboratory (IHL) have taken a different tack on citizen science: engaging disease-focused social networks. “Online communities not only provide a place for members to support each other, but also contain knowledge that can be mined for public health research, surveillance, and other health-related activities,” says Mandl.

At the end of April, the pair published the first results of a partnership with TuDiabetes.org, a social network focused on diabetes, in which they launched a “data-donation” drive encouraging community members to share their hemoglobin A1c status, a health metric used to measure diabetes control over a prolonged period of time.

TuAnalyze's aggregate map shows user-submitted A1c data on in a country- or state-level view, lighting up every time someone provides a data point.

The TuDiabetes community responded enthusiastically: Within three months, 17 percent of active members shared at least one A1c value using TuAnalyze, an application developed by Mandl and Weitzman and launched on the TuDiabetes site. The application allowed users to share their health data anonymously or publicly. The submitted data were aggregated and displayed on state- or country-level maps in near real-time.

“We were hoping to gauge the community’s willingness to share their personal data for public health surveillance,” Mandl notes, “and give them a tool that allowed them to securely share their data, all the while supporting socially-based encouragement and a sense of community activism.”

More than 30 percent of participants chose to share their personal A1c data publicly on their community profile. Interestingly, those who did so tended to have lower average A1c values, as did early adopters (those who signed on to TuAnalyze within the first two weeks of launch) and members who shared multiple A1c values versus only one.

Importantly, the average A1c values gathered through TuAnalyze nicely matched those reported in the Centers for Disease Control’s most recent National Health and Nutrition Examination Survey (NHANES). To Mandl and Weitzman, this reinforces the validity of their approach, and opens the door to a new way of conducing public health surveillance — one that gives researchers an opportunity to converse and build a rapport with communities.

“While they produce high-quality data, large, structured population-based reporting systems are not nimble, and provide no opportunity for interaction or feedback,” Weitzman says. “Science is changing and there is emerging an expectation and desire among participants for a continued research relationship and an opportunity to learn more about their own disease, for which online networks provide a platform.”