Scientists are now working on the application of Artificial Intelligence (AI) to psychiatry, with a mobile speech-based application that can categorize a patient's mental health status as well as or better than the human can. "We are in no way trying to replace clinicians," said Peter Holtz, a research professor at the Institute of Cognitive Science at the University of Colorado at Boulder. "But we believe we can create tools that will enable them to better monitor their patients," he added in a paper published in the Schizophrenia Bulletin.
Even when the patient does this for a casual visit, therapists base their diagnosis and treatment primarily on listening to the patient – an age-old method that can be subjective and unreliable, notes co-author of the paper, Brita Elvenwag, Cognitive Neuroscience in Tromsey, Norway.
“People are not perfect. They can be distracting, and sometimes there are subtle signs of speech and warning signs. Unfortunately, there is no objective blood test for mental health, "Elvevag said.
To make progress on this, Brita Elvevag, a cognitive neuroscientist at the University of Troms, Norway and Holtz has teamed up to develop a technology learning machine capable of detecting daily speech changes that herald mental decline.
For example, sentences that do not follow a logical pattern can be a critical symptom of schizophrenia. Changes in tone or pace can lead to mania or depression. Memory loss can also be a sign of cognitive and mental health problems.
"Language is a critical pathway for detecting a patient's mental states. Using mobile devices and AI, we can monitor patients daily and track these subtle changes, "Holtz said.
The new mobile app requires patients to answer a series of 5 to 10 minutes by talking on their phone. Among the various other tasks, they wondered about their emotional state, asked to tell a story, listen to a story and repeat it and gave a series of tests of touch and movement motor skills.
In a recent study, the team asked human doctors to listen to and evaluate the speech samples of 225 participants – half with severe psychiatric problems and half with healthy volunteers. They then compared those results with those of the machine learning system.
"We have found that computer models of AI can be at least as accurate as physicians," Holtz said.
If the application has detected a worrying change, it may prompt the patient's doctor to report it, the researchers noted.
If you have an interesting article / experience / case study for sharing, please contact us at [email protected]