Boston Children’s Hospital uses social media data for health research
Guest post by: Jared B. Hawkins, Ph.D., M.M.Sc., Research Associate, Boston Children’s Hospital & Harvard Medical School
Social media pervades the U.S. today. Take Twitter, for example. By the end of 2014, approximately one in five U.S. adults were active Twitter users. While the network remains most popular with adults under 50 years old, the last year saw a jump in tweeters 65 and older.
Despite growing privacy concerns, users of Twitter and other networks routinely talk about their health on social media. This has created a large and growing body of data and presented an opportunity to capture ‘digital phenotypes’ that provide tremendous insight into both individual and population health. These phenotypes let us:
- Identify individual patients suffering from acute or chronic disease and analyze their behavior over-time
- Monitor the health of a population by tracking the prevalence of infectious diseases (e.g., influenza)
Boston Children’s Hospital Computational Epidemiology Group has significant experience with—and has developed new technologies for—automated, informatics-based global health monitoring. This includes research based on social media data; we provided the scientific expertise behind Google Dengue Trends and advised Google on the development of Flu Trends.
In addition, we published some of the first research on using:
• News reports and social media to detect pandemics
• Facebook likes to understand obesity
• Twitter data to monitor emergence of infectious epidemics, like cholera in Haiti
• Twitter data to report news events in real-time, such as the Boston Marathon bombings
• Social media to measure public health sentiment towards vaccine acceptance
We’re going to present new research on sleep disorders and patient experience at Health Datapalooza’s Social Media Data Workshop on June 3. In collaboration with Merck, we’ve explored whether it is possible to identify patients who suffer from sleep disorders, and whether they differ from a control population, based on data from Twitter. Additionally, we’ll present methods for identifying tweets detailing patients’ perceptions of the quality of care they receive in U.S. hospitals and discuss the utility of this novel data stream.
We look forward to a spirited discussion about these and other directions in social media-based health research. For more information and full agenda, please see here.