Press Trust of IndiaJune 17, 2019 08:20:09 IST
Scientists who use artificial intelligence have discovered a hidden clue in people's language that accurately predicts whether they are likely to develop psychosis in the future.
The method of automatic learning quantifies more precisely the semantic richness of conversational language, a known indicator of psychosis.
The research, published in the journal npj schizophrenia, shows that an automated analysis of the two linguistic variables – more frequent use of words associated with sound and speaking with a low or unspecified semantic density – can predict whether a person at risk will develop a psychosis with an accuracy of 93%.
Even trained clinicians did not notice how people at risk for psychosis used more words associated with sound than the average, although abnormal hearing perception is a preclinical symptom.
"Trying to hear these subtleties in conversations with people, it's like trying to see the microscopic germs with the eyes," said Neguine Rezaii, who led the research at Emory University. in the USA.
"The automated technique we have developed is a very sensitive tool for detecting these hidden patterns.It's like a microscope to detect the warning signs of a psychosis," said Rezaii, currently at the # 39, Harvard University in the United States.
"We previously knew that the subtle traits of future psychosis were present in people's language, but we used machine learning to uncover hidden details on these points," said Phillip Wolff, a professor at the University of Toronto. Emory University.
The findings add to the evidence showing the potential of using machine learning to identify linguistic abnormalities associated with mental illness, said Elaine Walker, Professor of Emory.
The onset of schizophrenia and other psychotic disorders usually occurs in the early twenties, with warning signs – called prodromal syndrome – beginning around 17 years of age.
Approximately 25 to 30% of youth who meet the criteria for a prodromal syndrome will develop schizophrenia or other psychotic disorder.
With the aid of structured interviews and cognitive tests, trained clinicians can predict psychosis with an accuracy of nearly 80% in those who suffer from a prodromal syndrome.
Research in machine learning is one of many ongoing efforts to streamline diagnostic methods, identify new variables, and improve forecast accuracy. Currently, there is no cure for psychosis.
"If we can identify people at risk earlier and use preventative interventions, we may be able to address deficits," Walker said.
"There is good data showing that treatments such as cognitive behavioral therapy can delay the onset and perhaps even reduce the onset of psychosis," she said.
Tech2 is now on WhatsApp. Sign up for our WhatsApp services to take advantage of the latest technology and science. Just go to Tech2.com/Whatsapp and click on the Subscribe button.