Can artificial intelligence-led voice analysis help identify mental disorders?

This article is part of limited series about the potential of artificial intelligence to solve everyday problems.

Imagine a quick and easy test like measuring your temperature or blood pressure that could reliably identify an anxiety disorder or predict an impending depressive relapse.

Healthcare providers have many tools for assessing a patient’s physical condition, but there are no reliable biomarkers – objective indicators of externally observed medical conditions – for assessing mental health.

But some artificial intelligence researchers now believe that the sound of your voice could be the key to understanding your mental state – and AI is perfectly suited for detecting such changes, which are difficult, if not impossible, to detect otherwise. The result is a set of applications and online tools designed to monitor your mental status, as well as programs that provide real-time mental health assessment to telehealth providers and call centers.

Psychologists have long known that certain mental health problems can be detected not only by listening what the person says but how they say so, said Maria Espinola, a psychologist and assistant professor at the University of Cincinnati School of Medicine.

In depressed patients, said Dr. Espinola, “their speech is generally more monotonous, flatter and softer. They also have a reduced tone range and lower volume. They take more breaks. They stop more often. ”

Patients with anxiety feel more tension in their body, which can also change the way their voice sounds, she said. “They tend to talk faster. They have more difficulty breathing. ”

Today, these types of vocal characteristics are used by machine learning researchers to predict depression and anxiety, as well as other mental illnesses such as schizophrenia and post-traumatic stress disorder. The use of deep learning algorithms can reveal additional patterns and characteristics, such as those recorded in short voice recordings, which may not be obvious even to trained professionals.

“The technology we use now can highlight features that may be significant that even the human ear cannot comprehend,” said Kate Bentley, an assistant professor at Harvard Medical School and a clinical psychologist at Massachusetts General Hospital.

“There is a lot of excitement about finding biological or more objective indicators of psychiatric diagnoses that go beyond the more subjective forms of assessment traditionally used, such as interviews evaluated by clinicians or self-assessment measures,” she said. Other clues that researchers are monitoring include changes in activity levels, sleep patterns and social media data.

This technological advancement comes at a time when the need for mental health care is particularly acute: according to a report by the National Alliance on Mental Illness, one in five adults in the United States experienced mental illness in 2020. And the numbers continue to rise.

Although artificial intelligence technology cannot solve the lack of qualified mental health providers – there are not nearly enough of them to meet the country’s needs, said Dr. Bentley – There is hope that it could reduce barriers to obtaining a correct diagnosis, help clinicians identify patients who may be reluctant to seek care, and facilitate self-monitoring between visits.

“A lot can happen between deadlines, and technology can really offer us the potential to improve monitoring and evaluation in a more continuous way,” said Dr. Bentley.

To test this new technology, I started by downloading the Mental Fitness app from Sonde Health, a health technology company, to see if my weakness was a sign of something serious or just a clone. Described as “a product for monitoring mental fitness and keeping a diary based on voice”, the free application invited me to record my first application, a 30-second verbal entry in the diary, which would rank my mental health on a scale of 1 to 100.

A minute later, I had my result: not so great 52. The application warned “Pay attention”.

The app indicated that the level of liveliness in my voice was extremely low. Did I sound monotonous just because I was trying to speak softly? Should I listen to the suggestions of the application to improve my mental condition by going for a walk or relieving my space? (The first question may indicate one of the possible shortcomings of the application: as a consumer, it can be difficult to know why Your voice levels vary.)

Later, feeling nervous between interviews, I tested another voice analysis program, one that focused on detecting anxiety levels. The StressWaves Test is a free online tool from Cigna, a healthcare and insurance conglomerate, developed in collaboration with artificial intelligence expert Ellipsis Health to assess stress levels using 60-second recorded speech samples.

“What keeps you awake at night?” was the prompt of the website. After spending a minute recounting my persistent worries, the program took a snapshot of me and emailed me, “Your stress level is moderate.” Unlike the Sonde app, Cigna’s email didn’t offer helpful tips for self-improvement.

Other technologies add a potentially useful layer of human interaction, such as Kintsugi, a Berkeley-based California-based company that raised $ 20 million in Series A funding earlier this month.

Founded by Grace Chang and Rima Seiilova-Olson, who have teamed up to share previous experiences fighting for access to health care, Kintsugi is developing technology for telehealth providers and call centers that can help them identify patients who could benefit from further support.

Using Kintsugi’s voice analysis program, a nurse could be encouraged, for example, to take an extra minute to ask an annoyed parent with a cramped baby about his own condition.

One of the concerns regarding the development of these types of machine learning technologies is the issue of bias – ensuring that programs work equally for all patients, regardless of age, gender, ethnicity, nationality and other demographic criteria.

“For machine learning models to work well, you really need to have a very large, diverse and robust data set,” Ms. Chang said, noting that Kintsugi used voice recordings from around the world, in many different languages, to protect against this problem. especially.

Another big concern in this emerging area is privacy – especially voice data, which can be used to identify individuals, said Dr. Bentley.

Even when patients agree to be filmed, the question of consent is sometimes twofold. In addition to assessing a patient’s mental health, some voice analysis programs use recordings to develop and refine their own algorithms.

Another challenge, said Dr. Bentley, is a potential consumer distrust of machine learning and so-called black box algorithms, which work in ways that even the developers themselves cannot fully explain, especially which functions they use to predict.

“There is algorithm creation, and there is algorithm understanding,” said Dr. Alexander S. Young, interim director of the Semel Institute for Neuroscience and Human Behavior and the Department of Psychiatry at the University of California, Los Angeles, reiterating concerns many researchers have about AI and machine learning in general: that little, if any, human oversight is present during the training phase of the program.

For now, dr. Young remains cautiously optimistic about the potential of voice analysis technologies, especially as a tool for patients to monitor themselves.

“I believe you can model people’s mental health or roughly determine their mental health status in general,” he said. “People like to be able to monitor their own status, especially in chronic diseases.”

But before automated voice analysis technologies become commonplace, some require rigorous investigations into their accuracy.

“We really need more validation not only of voice technology, but also of AI and machine learning models built on other data streams,” said Dr. Bentley. “And we need to achieve that validation from large, well-designed representative studies.”

Until then, artificial intelligence-led voice analysis technology remains a promising but untested tool, one that could ultimately be an everyday method of measuring the temperature of our mental well-being.

Leave a Comment

Your email address will not be published.