Since 2013, Ieso has focused on depression and generalized anxiety disorder, and has used data-driven technology—of which NLP is a core part—to improve Recovery rate for these conditions Dramatically.According to Ieso, the recovery rate for depression in 2021 is 62%-by comparison National average 50%– and 73% of generalized anxiety disorders – compared to the national average of 58%.
Ieso said it focuses on anxiety and depression in part because they are the two most common conditions. But they also respond better to CBT than others, such as obsessive-compulsive disorder. It is not clear to what extent the clinic can expand its success, but it plans to start paying attention to more situations.
In theory, using artificial intelligence to monitor quality can free clinicians time to see more customers, because better treatment means less nonproductive treatment, although Ieso has not yet studied the direct effect of NLP on care efficiency. Influence.
“Now, with 1,000 hours of treatment time, we can treat 80 to 90 clients,” Freer said. “We are trying to move the needle and ask: Can you treat 200, 300, or even 400 clients in the same way? The number of hours of treatment?”
Unlike Ieso, Lyssn itself does not provide treatment. Instead, it provides its software to other clinics and universities in the UK and the US for quality control and training.
In the United States, Lyssn’s clients include California’s telemedicine opioid treatment program, which wants to monitor the quality of care provided by its providers. The company also cooperated with the University of Pennsylvania to use its technology to establish a CBT therapist in Philadelphia.
In the UK, Lyssn is working with three organizations, including Trent Psychological Therapies Service, an independent clinic that, like Ieso, is commissioned by the NHS to provide mental health services. Trent PTS is still trying out the software. Since the NLP model was established in the United States, the clinic had to work with Lyssn to enable it to recognize accents in the UK.
Dean Repper, Director of Clinical Services at Trent PTS, believes that the software can help therapists standardize best practices. “You would think that a therapist who has been doing this for years will get the best results,” he said. “But they don’t necessarily.” Repper likened it to driving: “When you learn to drive, you learn to do safe things,” he said. “But after a while, you will stop doing safe things and you may receive a fine for speeding.”
Improve, not replace
The focus of artificial intelligence is to improve human care, not to replace it. Short-term quick solutions cannot solve the lack of quality mental health care services. Solving this problem also requires reducing stigma, increasing funding and improving education. In particular, Blackwell refuted many claims against artificial intelligence. “There is a dangerous hype,” he said.
For example, there are a lot of topics about chatbot therapists and 24/7 monitoring of apps—often called the Fitbits of the brain. But most of this technology is somewhere between “a few years from now” and “never happen”.
“This has nothing to do with health apps and similar things,” Blackwell said. “Putting an app in someone’s hands and saying that it can treat their depression may only prevent them from asking for help.”
However, one problem that makes psychotherapy more evidence-based is that it means requiring therapists and clients to open up their private conversations. Would the therapists object to monitoring their professional performance in this way?
Repper expects some reluctance. “This technique is a challenge for therapists,” he said. “It’s like the first time they have someone in the room, transcribing everything they say.” First of all, Trent PTS only uses Lyssn software for students who want to be monitored. Repper believes that when these therapists are eligible, they may be monitored because they are used to it. It may be necessary to convince more experienced therapists of its benefits.
Immel said that the point is not to use technology as a stick, but as support. He used to be a therapist himself. He thinks that many people would welcome additional information. “It’s hard to face customers alone,” he said. “It’s really hard to improve when you sit in a private room with another person for 20 or 30 hours a week without feedback from colleagues.”
Freer agreed. At Ieso, therapists discuss AI-generated feedback with their supervisor. The idea is to let the therapists control their career development and show them what they are good at — things that other therapists can learn — rather than things they might want to work on that are not so good.
Ieso and Lyssn have only just begun to take this path, but only by digging a large enough data set can they discover the relevant knowledge of treatment.Atkins mentioned a Meta-analysis published in 2018 Approximately 1,000 hours of treatment were integrated without the help of artificial intelligence. “Lysn can handle this in one day,” he said. New research published by Ieso and Lyssn analyzes tens of thousands of conversations.
For example, in a paper published in JAMA Psychiatry in 2019, Ieso researchers Describes a deep learning NLP model that is trained to classify the therapist’s utterance In more than 90,000 hours of CBT courses with approximately 14,000 customers. The algorithm learned to distinguish whether different phrases and short parts of the conversation are examples of specific types of conversations based on CBT-such as checking the mood of the client, assigning and reviewing homework (the client practices the skills learned in the conversation), and discussing changes Methods, planning for the future, etc.-or conversations not related to CBT, such as general chat.