In a new study, researchers found that Facebook posts may help identify conditions such as diabetes, anxiety, depression, and psychosis in patients.
They found the language in posts could show diseases and help monitor symptoms.
The research was conducted by a team from Penn Medicine and Stony Brook University.
Social media posts are often about people’s lifestyle choices and experiences or how they’re feeling.
In the current study, the team examined whether these posts could provide information about disease management.
They used an automated data collection technique to analyze the entire Facebook post history of nearly 1,000 patients.
The patients agreed to have their electronic medical record data linked to their profiles.
The team then built three models to analyze their predictive power for the patients:
One model only analyzing the Facebook post language, the second model used demographics such as age and sex, and the third combined the two datasets.
The results showed that 21 different health conditions were predictable from Facebook alone.
Moreover, 10 of the conditions were better predicted through Facebook data than demographic information.
For example, “drink” and “bottle” in Facebook posts were shown to be more predictive of alcohol abuse.
The people that most often mentioned religious language like “God” or “pray” in their posts were 15 times more likely to have diabetes than those who used these terms the least.
Words expressing hostility, such as “dumb” and some expletives, might predict drug abuse and psychoses.
The findings show a link between language patterns and specific disease.
The team hopes the insights gleaned from their study could be used to better inform patients and providers about their health.
It is possible to develop an opt-in system for patients that could analyze their social media posts and provide extra information for clinicians to refine care delivery.
The lead author of the study is Raina Merchant, MD, MS, the director of Penn Medicine’s Center for Digital Health.
The study is published in PLOS ONE.
Copyright © 2019 Knowridge Science Report. All rights reserved.
