The language used on Facebook posts can help to identify cases of people with various diseases, such as diabetes, anxiety, depression and psychosis; according to new U.S. scientific research.
The way in which the posts are written by users, could be an indicator of disease and, with the consent of the patient, be tracked just like the physical symptoms.
The researchers analyzed over a period all the posts of about 1,000 individuals (who granted consent to correlate these data with their medical history) and created three disease forecasting models.
- One received exclusive information from Facebook,
- the second only demographics (age, gender, etc.)
- and the third combined both.
The results of the study published in the journal ”PLOS ONE”, indicate that all 21 cases of diseases that examined by the research, can be predicted from Facebook, even for 10 of them, Facebook was better predictor than demographics.
Indicatively, it seemed that the words ’bottle’ and ‘drink’ in the posts were associated with over-consumption of alcohol, while other words that reveal hostility, refer to drug abuse or psychosis. More unexpectedly, the reference to the words ”God” or ”prayer” is 15 times more likely to ’show’ people with diabetes.
”People’s choices of life, experiences, and feelings are often seen in social media. Such information can help provide additional information about the factors that aggravate or improve various diseases»,
said the Chief Researcher, Associate Professor Raina Mertsant, Director of the Medical Center for Digital Health at the University of Pennsylvania.
”There are many researches. so far, that show the relationship between the use of specific language terms and illnesses, such as the language predicts the depression. However, what is important is to find relations between these medical cases, something that can be a starting point for new developments in the field of artificial intelligence“,
said Andrew Swartz, Assistant Professor of computer science at the universities of Pennsylvania and Stoni Brooke.
A study last year by the same researchers (on which the new research is based) had shown that, via posts on social media, depression can be diagnosed even three months earlier than a clinical diagnosis.
Such data leave, according to Dr. Mertsant, room for creating a social media system that will give information about the health of the people to specialists, in order to ensure optimal care.
It remains to be seen if the implementation of this plan is feasible and whether the people will agree to share such information.