Depression is a very important public health issue. Fortunately, the means of detecting it are evolving greatly. For example, British scientists have recently developed an AI capable of defining the state of a person's mental health by analyzing, among other things, their profile on the social network Twitter.
According to the WHO, more than 300 million people would be affected by depression. This very common mental disorder is also the leading cause of disability in the world. On the treatment side, antidepressants have a good share. However, other means are emerging, such as transcranial alternating current stimulation (tACS). The symptoms of depression are numerous:loss of pleasure and interest, lowered self-esteem, feeling of uselessness, loss of appetite, dark thoughts, dejection or even constant and intense fatigue. What if some of these symptoms could directly be detected on the Internet ? This hypothesis is precisely that of researchers from the universities of Brunel and Leicester (United Kingdom) whose work was the subject of a first publication on the arXiv platform in 2019.
Researchers have set themselves the goal of detecting the state of mental health of Internet users on the social network Twitter. To do this, they have developed an artificial intelligence to analyze the profiles of users of the platform. Moreover, the tool has undergone training using a database of Twitter accounts. This data contained the history, the content of the messages, but also the times of publication, the number of followers as well as other additional information relating to mental health.
According to the results, it is about an accuracy rate of 88.39% regarding the detection of depression among Internet users. Abdul Sadka, one of the project's researchers, said in a TechXplore article on April 6, 2022:"We tested the algorithm on two large databases and compared our results to other techniques for detecting depression. In all cases, we managed to outperform existing tools in terms of classification accuracy. »
It is easy to think that the AI in question could be used to warn users of social networks that they might be subject to a mental disorder. However, British scientists believe that it is possible to envisage a use by the police services (or employers). The latter could investigate a person by observing their profile, their publications and their comments. On the other hand, this possibility could raise serious questions, in particular with regard to the respect for the private life of the people thus analyzed.
