Mental illnesses have become a modern day epidemic as many young people have succumbed to the various psychological challenges that they face every day. Some mental illnesses, like depression, are difficult to detect in people who suffer from it as a happy face can very easily hide what they are going through mentally and emotionally.
Now, diagnosing mental illnesses might
just be a little easier as new preliminary research done in America could
potentially detect conditions such as depression and diabetes using Facebook.
Researchers at the University of Pennsylvania and Stony Brook University carried out a new study that analysed the full Facebook post history of 999 patients, which totalled up to about 20 million words.
These patients agreed to have their Facebook profiles linked to their electronic medical record data.
The study looked into how Facebook posts could predict medical conditions across 21 broad categories using three different models that the researchers built.
Model 1: Analysed the language used in
patients’ Facebook posts
Model 2: Analysed patients’ demographics like age and gender
Model 3: Analysed data from Model 1 and Model 2 combined
In the results that were published in the journal PLOS ONE, all 21 medical conditions that were analysed in the study could be predicted just by Facebook posts alone.
What’s even more surprising is that the data gathered from Facebook was better able to predict 10 of the medical conditions than the demographic data. This includes illnesses like anxiety, depression, psychoses and even diabetes.
The researchers went about this study by using certain keywords on Facebook that were fairly intuitive at predicting certain diseases. For instance, words like “drink” and “bottle” were revealed to predict alcohol abuse in patients.
However, other links between words and diseases showed less of a prediction.