The findings, published in the journal PLOS ONE, showed that all of the 21 conditions included in the study could be predicted by Facebook alone. (Shutterstock/Alexey Boldin)
According to new preliminary US research, healthcare professionals could potentially use Facebook as a means of diagnosing conditions such as depression and diabetes.
Carried out by researchers at the University of Pennsylvania and Stony Brook University, the new study analyzed the complete Facebook post history of 999 patients, a total of around 20 million words, who had agreed to have their electronic medical record data linked to their Facebook profiles.
The researchers built three models to analyze whether the posts could predict medical conditions across 21 broad categories. One model analyzed the patient’s language used in their Facebook posts, another used demographics such as age and sex, and the last combined the two sets of data and analyzed both.
The findings, published in the journal PLOS ONE, showed that all of the 21 conditions included in the study could be predicted by Facebook alone. In fact, Facebook data was better at predicting 10 of the conditions than the demographic information, and was particularly effective at predicting diabetes and mental health conditions including anxiety, depression and psychoses.
Certain words used in Facebook appeared to be fairly intuitive at predicting whether users have certain diseases, for example “drink” and “bottle” were shown to be more predictive of alcohol abuse.
However, other associations between words and diseases were less obvious. For example, those who used religious language the most, using words like “God” or “pray” in their posts, were 15 times more likely to have diabetes than those who used these words the least.
Negative and hostile words such as “dumb,” and also some swear words, also predicted drug abuse and psychoses.
“Our digital language captures powerful aspects of our lives that are likely quite different from what is captured through traditional medical data,” said the study’s senior author Andrew Schwartz, PhD “Many studies have now shown a link between language patterns and specific disease, such as language predictive of depression or language that gives insights into whether someone is living with cancer. However, by looking across many medical conditions, we get a view of how conditions relate to each other, which can enable new applications of AI for medicine.”
“This work is early, but our hope is that the insights gleaned from these posts could be used to better inform patients and providers about their health,” added lead author Raina Merchant, MD, MS. “As social media posts are often about someone’s lifestyle choices and experiences or how they’re feeling, this information could provide additional information about disease management and exacerbation.”