Research News

Research News Image

Language Patterns May Predict Episodes Of Psychosis In At-Risk Individuals

Jun 25 2019
By studying the behaviors and experiences of at-risk individuals, researchers hope to identify signs of an oncoming episode of psychosis. Past research indicates that language patterns are one such sign. A new study used a machine learning system to analyze the speech patterns of 30 young people at risk of developing psychosis and 30,000 contributors to the social network site Reddit. The machine learning system was able to predict which at-risk individuals would develop psychosis in the following two years with 93% accuracy. This represents a major opportunity to intervene and potentially prevent poor outcomes for at-risk individuals. To learn more about this study, please visit the NIH website.