Artificial intelligence has been trained to diagnose post-traumatic stress disorder in copywriters. Article about it published • Boundaries in Psychiatry.
Jeff Savalha of the University of Alberta (Canada) and colleagues conducted a sentiment analysis of texts received during psychotherapeutic interviews. Sentiment analysis is a method of computational linguistics in which an algorithm counts emotionally charged areas, such as how many negative and positive thoughts there are in the text.
“We wanted to examine sentiment analysis of database texts and see if we could use it to distinguish the effects of post-traumatic stress disorder (PTSD) from emotional expressions alone,” says Savalha.
Psychotherapy interviews were received by a virtual character during a video conference, with 188 of its participants suffering from PTSD and 87 not. It turns out that the speech of people with PTSD is characterized by a large number of neutral or negative expressions. According to the authors, this is consistent with the scientific literature on the subject. According to her, some people with PTSD try to be neutral and suppress their emotions, and often try to talk less. Others, on the contrary, display overtly negative emotions.
As a result, the experts were able to use machine learning to create an algorithm that identifies people with PTSD with 80 percent accuracy. Scientists hope that in the future it will be possible to use it as an inexpensive diagnostic tool for telemedicine.