Trending Topics

Can health behaviors and conditions be predicted by social media?

Analysis of social media accounts has predictive clues about patients’ health behaviors and medical conditions

words_300.jpg

A slide from the American Heart Association’s Scientific Sessions 2015 presentation.

Photo by Greg Friese

ORLANDO, Fla. — Social media users share a vast amount of information about their everyday lives that can be used to understand their health and behaviors. Computer scientist Andrew Schwartz, Ph.D., Stony Brook University, described his efforts to connect social media big data with emergency care at the American Heart Association’s Scientific Sessions 2015.

New information, using social media network data, is readily available for understanding patient’s lives with legitimate statistical signals that is predictive of everyday health and medical conditions.

Memorable quotes on big data and emergency care
“Wearable devices are a great source of everyday information, but still relatively expensive and limited in the type of information they acquire.”

“People with STDs talk about body parts more often (on social media).”

“For predicting depression, your Facebook account is as accurate as other screening tools.”

Key takeaways on status updates and medical conditions
Here are three takeaways on the use of social media to understand patients’ everyday health behavior and health conditions.

1. Facebook updates and tweets predict health conditions
Researchers have developed models that can predict risk factors for health conditions, such as atherosclerotic coronary heart disease, by looking at status updates. Those models use semantically-related predictive words to create topics that are correlated to health conditions.

2. Patients talk about health conditions on social media
In a feasibility study, emergency department patients allowed researchers to examine their electronic medical record and private Facebook status updates. They found patients with a medical condition or diagnosis are likely to talk about their condition within their social network.

After the feasibility study, the researchers broadened their studies, comparing different medical conditions with word clouds or topics for people in the Philadelphia area. The additional research informed the correlation of predictive words to health conditions.

3. If you recognize an emergency on Facebook
Facebook status updates are predictive of health condition and behavior. If you see signs of a pending mental health emergency, such as expressions of severe depression and suicidal ideation in peer’s status updates, take action. Reach out and offer care.

“Peer support is the lifeline that never fails,” wrote Kelly Grayson, in the EMS1 column, “How EMS can be the voice of courage for one another”.

Greg Friese, MS, NRP, is a contributing editor at EMS1 and a public safety training and technology thought leader. His work translates incident analysis and research-to-practice insights into how-to guidance that supports clinical performance, operational readiness and workforce resilience. Friese writes frequently about practical technology adoption in public safety operations, including generative AI. He co-founded First Responder Wellness Week and co-hosts the Wellness Brief video series in the Lexipol Wellness app. Connect with Friese on LinkedIn or by email, greg@gregfriese.com.