Mining Social Media for Community Health

Mining Social Media

New Research from the Health Informatics Program

Tiffany Veinot sees great potential for technology to improve the health of populations everywhere. As director of the University of Michigan's Master of Health Informatics program, a partnership of the School of Public Health, the School of Information, and the School of Medicine's Department of Learning Health Sciences, Veinot leads a cross-campus educational initiative bringing technology to bear on human health outcomes.

One of the fastest-growing fields in the nation, health informatics uses innovative information technology to improve health and healthcare. Health informatics is a highly interdisciplinary field, and Michigan's program combines the School of Public Health's population health perspectives and world-class expertise in health policy and health behavior with the School of Information's expertise in human-centered design, social computing, and the development and evaluation of novel information resources, as well as the School of Medicine's emphasis on socio-technical infrastructures to build a learning health system.

The project has explored tweets concerning nutrition, physical activity, and sedentary behavior at the census-tract level and correlated census tract–based measures with offline characteristics such as neighborhood disadvantage, fast-food store density, and mortality rates related to specific chronic conditions.

Veinot is especially interested in "social media's potential to characterize the impact of location—the places people live—on their health," she says. As part of the Social Sciences Annual Institute and Mcubed-funded project "A 'Big Data' Approach to Understanding Neighborhood Effects in Chronic Illness Disparities," Veinot and her team of collaborators, including Michigan Public Health's Philippa Clarke and Veronica Berrocal, use Twitter to glimpse the most granular levels of neighborhood resources and attitudes toward health in Southeast Michigan populations.

The project has explored tweets concerning nutrition, physical activity, and sedentary behavior at the census-tract level and correlated census tract–based measures with offline characteristics such as neighborhood disadvantage, fast-food store density, and mortality rates related to specific chronic conditions.

As health systems across the US evolve their digital infrastructures, health informatics professionals will play an increasingly crucial role in the integration of information technology and health care. Veinot's team is now working to establish and validate their social media–based measures, a remarkably complex process considering how unstructured social media signals like natural language tweets are.

Veinot's ultimate goal is "an improved granularity and currency of neighborhood health information that upstream decision makers can deploy to help them make decisions that improve the health of marginalized communities." Though she acknowledges the field is in its infancy and has "a number of steps before we get there," Veinot is inspired by the promise of creating novel measures of health at a community level.

Tags