Researchers have developed a speech-based mobile app that uses artificial intelligence to categorize a patient’s mental health status, an advance that may lead to a tool to assist psychiatrists in diagnosing mental illnesses.
The study, published in the journal Schizophrenia Bulletin, noted that many people in remote areas do not have access to psychiatrists or psychologists, and others can’t afford to see a clinician frequently.
The researchers, including those from the University of Colorado at Boulder in the US, said therapists base their treatment plan largely on listening to a patient talk, which they said was an old, subjective and unreliable method.
They developed a machine learning technology that can detect day-to-day changes in a speech, which hints at mental health decline.
As an example, they said, sentences that don’t follow a logical pattern can be a critical symptom in schizophrenia.
Shifts in tone or pace may suggest mania or depression, and memory loss can be a sign of both cognitive and mental health problems, the researchers said.
“Language is a critical pathway to detecting patient mental states,” said study co-author Peter Foltz from the University of Colorado at Boulder.
“Using mobile devices and AI, we are able to track patients daily and monitor these subtle changes,” he added.
The study noted that the new mobile app asks patients a 5- to 10-minute series of questions which they can answer by talking into their phone.
The patients are asked about their emotional state, or to tell a short story, or to listen to a story and repeat it.
The app also gives them a series of touch-and-swipe motor skills tests.
It assesses the speech samples, compares them to previous samples by the same patient and the broader population, and rates the patient’s mental state.
The researchers also asked doctors to listen to and assess speech samples of 225 participants – half with severe psychiatric issues and half healthy volunteers – in rural Louisiana in the US and Northern Norway.
When they compared the results to those of the machine learning system, they found that the computer’s AI models can be at least as accurate as the clinicians.
The researchers called for larger studies with the app to prove its efficacy and earn public trust.
“The mystery around AI does not nurture trustworthiness, which is critical when applying medical technology,” they said.
“Rather than looking for machine learning models to become the ultimate decision-maker in medicine, we should leverage the things that machines do well that are distinct from what humans do well,” the researchers added.
