The advances in AI has enabled computers to assist doctors in detecting diseases and help keep a check on patient health remotely. Now, researchers from the University of Colorado Boulder (CU Boulder) are working to leverage ML to psychiatry using a speech-based mobile app. This app can categorize patients’ mental status better than humans.
Peter Foltz, a research professor at the Institute of Cognitive Science says – “We are not in any way trying to replace clinicians. But we do believe we can create tools that will allow them to better monitor their patients.” Notably, he is also the co-author of a new paper in Schizophrenia Bulletin that illustrates the promise and potential pitfalls of artificial intelligence in psychiatry.
According to a study, around one in every five citizens of the US live with a mental illness and many of them reside in the far-off areas where access to psychiatrists or psychologists is quite less. And some can’t even afford to consult a clinician frequently, don’t have time or can’t get in to see one.
The co-author Brita Elvevåg of the paper, a cognitive neuroscientist at the University of Tromsø, Norway notes that even if patients consult doctors their therapies are based on the age-old method listening to a patient talk, which is subjective and unreliable. He says that humans are not perfect and can get distracted and miss out on subtle speech cues. He also says – “Unfortunately, there is no objective blood test for mental health.”
To overcome the shortcomings in the treatment process, both the experts teamed up to develop machine learning system that will be able to detect day-to-day changes in speech. This graph can hint at mental health decline.
Foltz states – “Language is a critical pathway to detecting patient mental states. Using mobile devices and AI, we are able to track patients daily and monitor these subtle changes.”
The mobile app used in the development process tends to ask patients to answer a 5- to 10-minute series of questions by talking to their phone. In this test, they’re asked about their emotional state, asked to tell a short story, listen to a story and repeat it and given a series of touch-and-swipe motor skills tests.
Peter Foltz and Brita Elvevåg collaborated with Chelsea Chandler, a computer science graduate student at CU Boulder, and other colleagues, to develop an AI system that is capable of assessing the speech samples collected from app, compares them to previous samples by the same patient. After the assessment, the system rates the patient’s mental state.
The team also compared the results from human clinicians, who assessed those speech samples of 225 participants, with that presented by the AI system. They found that the computer’s AI models can be at least as accurate as clinicians.
Peter and his team-mates dream of a day when AI systems they’re developing for psychiatry could share the same space with a therapist and a patient to provide additional data-driven insight for severely mentally ill people. In that case, if the app detects a worrisome change, it could notify the patient’s doctor to check in immediately.