Skip to main content

Vagueness of Speech, Talking About Sounds Predict Psychosis

July 03, 2019

Talk about voices and sounds combined with low semantic density, or vagueness of speech, predicted conversion to psychosis with 93% accuracy in at-risk individuals, according to a machine learning study published in the journal npj Schizophrenia.

“It was previously known that subtle features of future psychosis are present in people's language, but we've used machine learning to actually uncover hidden details about those features,” said senior author Phillip Wolff, PhD, a professor of psychology at Emory University, Atlanta, Georgia.

For the study, researchers used machine learning to compare conversational norms gleaned from 30,000 contributors on social media with the speech of 40 young people at risk of psychosis from the North American Prodrome Longitudinal Study.

Automated analysis showed that 2 language variables—speaking with low semantic density, otherwise known as poverty of content or vagueness, as well as using words associated with voices and sounds—were more common in at-risk individuals who later transitioned to psychosis.

Second-Generation Antipsychotics Recommended for Negative Symptoms

The two variables together predicted conversion with 93% accuracy in the training group and 90% accuracy in holdout datasets, according to the study. In comparison, trained clinicians using structured interviews and cognitive tests can predict psychosis with about 80% accuracy in patients with prodromal syndrome.

“In the clinical realm, we often lack precision,” said study first author Neguine Rezaii, MD, a fellow at Harvard Medical School in Boston, Massachusetts. “We need more quantified, objective ways to measure subtle variables, such as those hidden within language usage.”

Although there is currently no cure for psychosis, early intervention can make a difference.

“If we can identify individuals who are at risk earlier and use preventive interventions, we might be able to reverse the deficits,” said study coauthor Elaine Walker, PhD, psychology and neuroscience professor at Emory. “There are good data showing that treatments like cognitive behavioral therapy can delay onset, and perhaps even reduce the occurrence of psychosis.”

—Jolynn Tumolo

References

Rezaii N, Walker E, Wolff P. A machine learning approach to predicting psychosis using semantic density and latent content analysis. npj Schizophrenia. 2019 Jun 13;5(1):9.

Clark C. The whisper of schizophrenia: Machine learning finds 'sound' words predict psychosis [press release]. Atlanta, Georgia: Emory University; June 14, 2019.

Back to Top