Broadcast: News items

Automated speech analysis may detect psychosis

An automated speech analysis program demonstrated 100% accuracy in identifying whether at-risk young people went on to develop psychosis over the next two years. 

This study involved a small sample of 34 participants age 14 to 27 at clinical risk for psychosis. Participants were asked to describe a subjective experience in an interview and the computer program analysed their narrative; focusing on semantics, structure, coherence, phrase length and determiner words to link phrases. 

After 2.5 years, 5 participants developed psychosis and 29 did not. The computer program correctly differentiated these patients. This speech analysis computer program may increase predictive power beyond expert clinical opinion. The program identified that breaks in the flow of meaning, shorter phrases and less elaboration were key determinate features.

Clinicians do assess disorganised speech on the basis of clinical observation, but this objective tool may provide complementary evidence for patient assessment. However, it must be acknowledged that these results need to be replicated in larger samples.


Bedi G, Carrillo F, Cecchi GA, et al. Automated analysis of free speech predicts psychosis onset in high-risk youths. npj Schizophrenia. 2015 August 26. [Epub ahead of print].



By: Abigail Christine Wright
Last updated: Wednesday, 9 September 2015