ARTIFICIAL INTELLIGENCE PREDICTS PSYCHOSIS

by Ben Daviss

Psychosis – the mental illness in which people hear voices and otherwise hallucinate – often pops up in early adulthood and is hard to predict. Psychologists have to rely on conversations and paper-and-pencil tests to make their best guesses about whether a person will break with reality. 

Artificial intelligence now offers a more precise technique.

Researchers at Emory and Harvard universities used an AI program to find patterns in the speech of people who become psychotic. The program found that persons slipping into psychosis begin to show less richness in their spoken language and to use more words that describe or relate to sound.

The predictive program has been accurate in about 90 percent of people tested.

TRENDPOST: This new diagnostic tool not only may help catch people before they drop off the edge of sanity but also shed light on how psychosis develops – a step toward earlier and better treatment and, ultimately, to prevention.