For Darin Carver, the assistant clinical director for Weber Human Services, the often overlooked crisis in mental health treatment is not just one of access. It’s that the care people do get often doesn’t help.
There are proven techniques for helping people with things like depression, post-traumatic stress disorder and substance abuse. But in Utah, only about 43% of adults receiving mental health treatment improve and recover, according to data from the Utah Division of Substance Abuse and Mental Health.
While Carver’s organization tops the list of public behavioral health centers in urban counties, with about a 50% recovery rate, he said behavioral health overall is not where it needs to be.
“The quality is being overlooked,” he said. “It’s time to improve that, versus just to have conversations about access.”
Carver said part of the problem is clinicians can get stuck on the treatments they know best, even if they might not be the best option for a particular client. And historically, therapists have had few opportunities to get feedback on their approach in order to adjust.
That’s starting to change, thanks in part to growing research around the use of Artificial Intelligence in therapy. Some of it is being pioneered in Utah.
About three years ago, Carver’s organization was approached by Lyssn, a company that’s developed machine learning software to transcribe and analyze recordings of therapy sessions. It can measure things like time spent listening and how empathetic therapists are during treatment.
Lyssn co-founder Zac Imel, who’s also a counseling psychology professor at the University of Utah, said we’re likely a long way off from robot therapists, but machine learning software can for the first time scale quality evaluation in psychotherapy.
“This is feedback you could get from a human rater, but it's very expensive and slow,” he said. “We can give it to you within five minutes at the end of your session.”
Imel said psychotherapy experts had to first teach the software what to look for, then it was able to comb through thousands of hours of recordings and spit out reports that catalogue important indicators, analyzing everything from the words a therapist uses to the sound of their voice and the pauses they use.
Carver said the software is a tool he’s used as part of a larger effort to improve treatment. The data starts to provide a picture of how the organization is doing overall, as well as the ability to pinpoint specific moments in a therapy session and areas clinicians may need more practice in.
“That's a big change in the mental health field,” he said.
Imel said there are challenges with the software. Sometimes the machine just doesn't pick up on everything and the data it’s fed isn't always perfect, so there are multiple potential sources of error.
He said that’s where they go back to the clients and providers using the technology for feedback on how it’s working. Carver said it’s not clear yet precisely how much Lyssn has helped his organization, but he’s confident it’s having a positive impact.
“There's lots of imperfections in the system, but I do think it's a lot better than nothing, which is kind of what we've had historically,” Imel said.
