Psychosis Prediction Using Speech Patterns


Linguistic software to predict and diagnose psychosis.

Key Benefits

  • New method of diagnosing psychosis.
  • Very high accuracy rate (86.7%).

Market Summary

Psychotic disorders are mental illnesses characterized by delusions, hallucinations, and disorganized thinking and speaking. Schizophrenia is an example of a psychotic disorder. Such disorders cause a detachment from reality. Despite years of research, the biological causes of psychotic disorders and methods of diagnosing or predicting these disorders are unknown. The language of patients with psychosis has been known to contain indicators of mental illness. It is widely observed that poverty of content is a common feature of the language spoken by those with psychosis. However, the measurement of poverty of content is subject to the judgment of the clinician and not reliable.

Technical Summary

Emory researchers have developed a computational technique that can measure semantic density in a person’s speech. The inventors show that this method, in combination with automatic analyses of the latent semantic themes, can predict the future emergence of psychosis at a relatively high level of accuracy (86.7%).

Developmental Stage

The set of programs needed for running this analysis are now available.

Publication: Rezaii, N. et al. (2019). Npj Schizophrenia. 5(1), 9.

Patent Information

Tech ID: 18180
Published: 1/14/2020