Researchers have created a speech-based mobile app that makes use of artificial intelligence to categorize a person’s mental health condition, an advance that could result in a device to aid psychiatrists in diagnosing mental illnesses.
The study, released in the journal Schizophrenia Bulletin, stated that lots of individuals in remote areas don’t have access to psychologists or psychiatrists, and others cannot afford to see a clinician regularly. The researchers, together with those from the Faculty of Colorado at Boulder in the US, stated therapists base their treatment program mostly on listening to an individual discuss which they said was a well used, unreliable and subjective method.
They created a machine learning technology that can identify daily variations in the speech that hints at brain health decline. As a good example, they stated, sentences that do not adhere to a logical pattern could be a serious indicator of schizophrenia.
Shifts in pace or tone may suggest depression or mania, and memory loss could be an indication of both mental and cognitive health problems, the researchers said. Language is a crucial pathway to detecting patient brain states,” said study co-author Peter Foltz from the University of Colorado at Boulder.
“Using mobile products and AI, we’re in a position to observe patients every day and keep track of these subtle changes,” he included.
The study observed that the new mobile app asks the patient a 5-10 minute series of questions which they are able to reply by speaking into the phone.
The individuals are asked about the emotional state, or perhaps to tell a brief story, or perhaps to pay attention to a story and repeat it. The app also provides them a sequence of touch-and-swipe motor abilities tests.
It assesses the speech samples, compares these to previous samples by the same patient and the broader public, as well as rates the person’s psychological state. The scientists also requested medical professionals to tune in to and evaluate speech samples of 225 participants – one half with serious psychiatric problems and half healthful volunteers – in rural Louisiana within the US and Northern Norway.
When they compared the outcomes to those of the machine learning process, they discovered that the computer’s AI camcorder models may be at least as precise as the clinicians.
The researchers called for larger experiments with the app to confirm its efficacy and get public trust.
“The mystery around AI doesn’t nurture trustworthiness, which is crucial when implementing healthcare technology,” they said.
“Rather than searching for machine learning models to turn into the decision-maker in medicine, we have to leverage the items that models do very well which are unique from what people do well,” the researchers included.