AI LISTENS FOR EARLY SIGNS OF MENTAL ILLNESS

by Bennett Daviss
More than one in ten people worldwide are afflicted with a mental illness, yet often the signs are too subtle to detect until the illness is on full display.
But those signs aren’t too subtle for artificial intelligence.
At Vanderbilt University Medical Center, scientists created machine learning software to analyze a person’s age, gender, medication, and past diagnoses gathered from hospital records to predict that person’s risk of suicide.
In testing the program on more than 5,000 patients admitted to hospitals for harming themselves or attempting suicide, the program succeeded in 84 percent of the cases in predicting who would attempt suicide within a week and 80 percent accurate in predicting suicide attempts within the following year.
Apps are being tested, especially for teens, that monitor changes in cell phone behavior such as typing speed, tone of voice, and word choices to determine depression, anxiety, and other emotional turbulence. The phone could then send a warning to a parent or other adult.
TRENDPOST: Before 2030, physicians will be able to install artificial intelligence programs in their exam rooms that will listen in on conversations between doctor and patient. The software will mark changes in tone, subject, and other content and alert the doctor to early signs of mental illness or emotional disturbance.