Researchers predict individuals’ medical diagnoses from language posted on social media

There was once a time in life where the only way to talk to your neighbour was going to their house, a time where it was impossible to send any celebrity a direct message and they would reply but those times are gone. We now live in times where we can simply connect with the world just by opening an app.

These times are good because they can bring the world closer together and give people a chance to connect and better understand the next individual or the next ethic group.

However, these times, the social media times can hide and reveal a lot about people. Some people appear happy when they are really depressed and vice versa.

A group of researchers took it upon themselves to study whether medical conditions across 21 broad categories were predictable from social media content.

In a study titled ‘Evaluating the predictability of medical conditions from social media posts’ the researchers were keen to answer two questions they found important:

  1. Is it possible to predict individuals’ medical diagnoses from language posted on social media?
  2. Is it possible to identify specific markers of disease from social media posts?

To answer these questions they analysed 949,530 Facebook status updates containing 20,248,122 words across 999 participants whose posts contained at least 500 words. It sounds a bit crazy right. What they did was to evaluate whether the consenting patients’ Facebook posts could be used to predict diagnoses evident in their electronic medical records (EMR).

These participants consented to taking part in the study and diagnoses from their electronic medical records were collected and grouped into 21 categories according to the Elixhauser Comorbidity Index.

The results show that all 21 medical condition categories were predictable from Facebook language beyond chance, 18 categories were better predicted from a combination of demographics and Facebook language than by demographics alone, and 10 categories were better predicted by Facebook language than by the standard demographic factors (age, sex, and race).

This study tells us a lot about people’s personalities, mental state, and health behaviours as they subconsciously reflect them through their choice of words on social media, especially Facebook as referred to in this study.

Alcohol abuse was marked by topics mentioning drink, drunk, bottle. Issues most associated with depression suggested physical symptoms (e.g. stomach, head, hurt) and emotional distress (e.g. pain, crying, tears).

The researchers admit that this study has several limitations, therefore they suggests that Future studies should compare the difference in health related information disclosed by users of different demographic populations and on other social media platforms such as Twitter.

About Mduduzi Mbiza 110 Articles
Mduduzi Mbiza is a creator. Author of the book, ‘Human Education: The Voyage of Discovery’.