Your Facebook profile might be able to predict some medical conditions.

Researchers from Penn Medicine and Stony Brook University conducted a study to determine if social media posts can be indicators of 21 health conditions, including diabetes, anxiety, depression and psychosis.

To do so, they examined the Facebook history of nearly 1,000 patients who agreed to have their electronic medical record data linked to their profiles. Researchers then built three models to assess the subjects’ “predictive power.” One evaluated their Facebook post language; the second looked at demographics, such as age and sex; and the last one combined the first two data sets.

After analyzing the results, they found all 21 conditions were predictable from the social media network alone. For instance, those who used “drink” and “bottle” had an increased risk of alcohol abuse. And terms expressing hostility, such as expletives, were more predictive of drug abuse and psychoses.

“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,” lead author Raina Merchant said. “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.”