How Your Facebook Posts Can Reveal Your Medical Status
Researchers were able to correctly guess the health status of some people by analyzing their Facebook posts, according to a new study . The people involved have given their consent to the study and the results are confidential, but let this be a warning: you can potentially tell the world a lot more than you think.
In the study, the researchers asked people at the clinic – mostly African-American women – if they could analyze their social media posts and compare them with their medical records . They then trained an algorithm to look at the words people use in their messages and notice which words and word combinations are most likely associated with certain diseases. They trained algorithms on some of the data and then applied the algorithms to the rest of the data to try and guess the health status of the Facebook posters.
For some conditions, such as depression, Facebook posts were no better predictor of the condition than simply looking at demographic information from a poster such as age, gender, and race. But for others, Facebook’s data was more accurate, including diabetes, anxiety, and psychosis.
So far, this is just academic research; Facebook does not yet have access to your medical records and does not analyze your symptoms. (Facebook was not involved in the study, but one of the researchers does have ties to Microsoft.) But the study serves as a somewhat eerie reminder that we talk about ourselves more than we think.
The control words weren’t always obvious. “Topics expressing hostility (eg people, dumbass, bullies, bitches ) were the predominant marker of drug abuse as well as marked psychosis,” the authors write. “Topics associated with depression have included somatization (eg, stomach, head, pain ) and emotional distress (eg, pain, crying, tears ).”
This type of algorithm – like any other algorithm – can also catch some wrong things. Some markers of diabetes in this study were religious words such as god, family, and prayer . Maybe people with diabetes are starting to think a lot about God, or, more likely, it’s just a coincidence related to the age of the patients and other demographics. The study was small – only 999 people had enough Facebook posts to analyze – so if the same approach were applied around the world, the results might be different.
Since we all pay a little more attention to the information we share online, this research is another wake-up call that we may be sharing more than we think. In the future, this analysis could be used for good – perhaps the algorithm could, with your consent, inform you or your family that you may have a health condition that is worth checking. But for now, it’s also a reminder to consider what you post and what you post if you don’t want your social media profile to be overly explicit.