Twitter use could predict loneliness, mental health issues

By Dr. Liji Thomas, MDNov 4 2019

For all the rush and busyness of daily life, it gets to be very lonely, for at least 20% of people in the US. And since loneliness is a potent ingredient in the mix that produces heart disease, stroke, dementia and depression – among other health issues – it deserves a place in public health policy.

A new study published in BMJ Open on November 4, 2019, says that by analyzing tweets and other content posted on the popular social media app called Twitter, it is possible to associate certain types of post with the risk of loneliness. Tweeters with loneliness issues post more about relationship problems, substance abuse, sleep issues, and mental health. This could lead to a more rapid identification of and support to lonely users even if they don’t say so in so many words.

Image Credit: Burdun Iliya / Shutterstock

Researcher Sharath Chandra Guntuku says loneliness can be “a slow killer”, causing medical problems that take many years to show themselves. Hence the need to “identify lonely individuals and intervene before the health conditions associated with the themes we found begin to unfold,” which “could be very powerful and have long-lasting effects on public health.”

The study and its findings

The study hinged on finding themes and words to serve as markers of loneliness-associated content, and the authors say they have made a start on building a method to predict loneliness. The data was taken from Pennsylvania Twitter users whose accounts could be read by anyone. Analysis of these timelines led to the identification of over 6200 accounts which used words like “lonely” or “alone” over the study period of 2012 to 2016.

They then extracted the data from the whole Twitter account of each of these users. This was compared to a matched group of accounts in which these marker words did not occur. The researchers found that users who were lonely tweeted twice as often, with a markedly higher frequency at night.

When linguistic models were applied to the tweeted content, the analysis of the “lonely” group brought out a high incidence of words associated to anxiety, depression and anger compared to the control group. In addition, the first group used words like “no one to” or “want someone” to express relationship needs;  showed a lack of emotional control by phrases such as “I just wanna” or ”I can’t” or foul language; and their tendency to abuse substances, by tweeting “smoke”, “drunk” or “weed”.

The non-lonely group of Twitter users had a greater level of social connectedness. They often included names in their chats and had a greater number of conversations. It all seems to prove that we need to be nice to others, and that we need others, to lead a fulfilled life, quite apart from the fact that it helps us to fulfil others’ needs.

Implications

Apart from validating older concepts that substance use, emotional indiscipline, and inability to sustain meaningful relationships plays a key role in causing a lonely unfulfilled existence, the research also shows that the bidirectionality of this association needs to be explored. In other words, it is quite possible that not only does loneliness cause anger, but angry individuals are indeed much more likely to be lonely.

Going on from the findings, senior study author Raina Merchant says loneliness can be helped using many methods. The exact technique used will depend on the individual and the circumstances. She explains, “It's clear that there isn't a one-size-fits-all model. Some interventions include buddy systems, peer-to-peer networks, therapy, and skill development for navigating day-to-day interactions with others.” In other words, some people just need to be connected to other people because they aren’t very good at making new friends. But others don’t know how to get along with people, lacking the necessary social skills like regulating one’s emotions, showing appropriate behavior, and exercising self-control and good judgment. In that case, therapy will include sessions on modeling and practicing good social interactions. A few will have historical issues and will need counseling to deal with such traumas,  and move forward to embrace new friendships, hopefully after correcting personal factors that may have helped to spoil previous relationships.

The future

The researchers now intend to develop their model of loneliness prediction to include many more different aspects of loneliness. They have already developed a working predictive tool which accurately identified loneliness in a group of patients who volunteered to share their Twitter data and also filled out a recognized loneliness questionnaire. They hope that this will lead to a better tool that can be used to find lonely hospitalized patients, and then to develop effective interventions that will help them, as well as their families, friends and other supportive people in their lives.