EDMONTON, Alberta — A text message may be able to reveal if someone is dealing with post-traumatic stress disorder (PTSD), a new study finds. Researchers from the University of Alberta say a machine learning program — a form of artificial intelligence — is capable of “reading between the lines” to find potential warning signs in the way people write.
The team believes this program could become an inexpensive tool that helps mental health professionals detect and diagnose cases of PTSD or other disorders. Psychiatry PhD candidate Jeff Sawalha performed a sentiment analysis of texts using a dataset created by Jonathan Gratch from USC’s Institute for Creative Technologies.
Study authors explain that a sentiment analysis takes a large amount of data and categorizes it. In this case, the model took a massive amount of texts and sorted them according to positive and negative thoughts.
“We wanted to strictly look at the sentiment analysis from this dataset to see if we could properly identify individuals with PTSD just using the emotional content of these interviews,” Sawalha says in a university release.
PTSD texts are neutral or numb
The text sampling came from 250 semi-structured interviews conducted by an artificial interviewer (Ellie) who spoke with real participants using video conferencing calls. Eighty-seven people had PTSD while the other 188 did not.
From their text responses, the team was able to identify people with PTSD through their scores reflecting how often their words displayed neutral or negative thoughts.
“This is in line with a lot of the literature around emotion and PTSD. Some people tend to be neutral, numbing their emotions and maybe not saying too much. And then there are others who express their negative emotions,” Sawalha says.
Study authors note that this process isn’t black and white. For example, a phrase like “I didn’t hate that” could be confusing for the algorithm. Despite that, the machine learning system was able to detect PTSD patients with 80 percent accuracy.
“Text data is so ubiquitous, it’s so available, you have so much of it,” Sawalha continues. “From a machine learning perspective, with this much data, it may be better able to learn some of the intricate patterns that help differentiate people who have a particular mental illness.”
Finding different and cheaper ways of detecting mental disorders
The team is planning to integrate other types of data, including speech patterns and human motions, which they say may help the system spot mental health disorders better. Moreover, signs of neurological conditions like Alzheimer’s disease are detectable through a person’s ability to speak.
“Unlike an MRI that takes an experienced person to look at it, this is something people can do themselves. I think that’s the direction medicine is probably going, toward more screening tools,” says Russ Greiner, a professor in the Department of Computing Science.
“Having tools like this going forward could be beneficial in a post-pandemic world,” Sawalha concludes.
The study is published in the journal Frontiers in Psychiatry.