Apply Artificial Intelligence and social media to detect eating disorders

Anorexia and bulimia are probably the two best known ways in which Eating Disorders (TCA). Today, November 30, is celebrated World ADD Day to make the population aware of a problem that affects a very high percentage of our society; And the question is how teenagers feel about their bodies, the answer is that almost half want to lose weight. That's how he finds it the survey published yesterday by the Association Against Anorexia and Bulimia (ACAB), which points out that 47% of girls and 21% of boys want to lose weight, an increase of 15 and 6 points respectively compared to the previous year.

There is a consensus among experts that it is very important deploy prevention strategies to reduce the onset and development of ADD. They also agree that the social media and websites advocating anorexia and bulimia they are a growing danger and need to be identified. And it is at this intersection, between this type of disorder and content on the web, where the STOP project (Suicide Prevention in Social Platforms).

This project, led by the researcher Ana Freire of UPF School of Management, with the participation of 7 research nodes, including the Parc Taulí University Hospital i theParc Taulí Research and Innovation Institute (I3PT), was born with the goal of searching for and analyzing patterns of suicidal behavior by applying Artificial Intelligence (AI) to social media. Specifically, a group of researchers, engineers, psychologists and psychiatrists developed AI algorithms that, through the analysis of texts, images and activity on Twitter, detect patterns of suicidal behavior.

Currently, eating disorders are also analyzed using this technology. The clinical psychologist at Parc Taulí Nadia Sanz, who actively participates in the STOP project, explains that thanks to the work of a multidisciplinary team has been achieved detect people at risk for these disorders. “The team of psychologists and psychiatrists we collected and filtered tweets that they contain keywords such as “BMI”, “weight” or “height”, and we have coded them clinically according to whether they are at risk for ADD; later a group of engineers have created some algorithms that detect profiles of people at risk according to their patterns of behavior ", explains Sanz, who emphasizes that all this information is collected and managed anonymously, so that at no time is the identity of the users known to respect their privacy. The ultimate goal is to create a campaign that reaches users which fit within the detected pattern for people at risk for ADD. These young women are introduced to the Telephone of Hope so they can contact them and ask for help.

Sanz stresses that it is very important detect disorders much more quickly in order to offer help at a more early stage. The pandemic, he says, has been very negative for people with or at risk for ADD and has been seen as they have increased the cases very significantly. "The STOP project, thanks to the work of this multidisciplinary team, is helping to provide these girls with a much-needed help service at the moment."