At NatCon, Lyssn hosted a panel focused on a critical challenge in behavioral health: access to effective treatment for serious mental illness (SMI), particularly psychosis.
Today, fewer than 1% of behavioral health practitioners are trained in interventions for psychosis. That gap has real consequences, limiting access to care for clients who need it most.
Our panel featured Brian Allender, MD, Chief Medical Officer at King County; Zac Imel, PhD, Chief Science Officer at Lyssn; Sarah Kopelovich, PhD, ABPP, Associate Professor at the University of Washington; and Christina Soma, PhD, Research Scientist and Clinical Implementation Consultant at Lyssn. Together, they explored whether AI can play a meaningful role in closing that gap and what it takes to do so responsibly and effectively.
The discussion was grounded in recent research conducted in partnership between the University of Washington and Lyssn. The study evaluated a new approach to training clinicians in SMI interventions using AI-enhanced, practice-based learning.
Rather than relying solely on traditional training methods, clinicians engaged in shorter, focused learning sessions paired with immediate, AI-generated feedback on their practice. The results showed measurable improvements in both provider skill and client recovery outcomes compared to standard training approaches.
This represents an important step forward, not just in how training is delivered, but in how quickly and consistently clinicians can build and apply these skills in real-world settings.
Key takeaways from the panel included:
- AI Is a Tool, Not a Replacement
AI should be positioned as a tool to help human providers deliver better care — not to replace them. Setting this expectation clearly is critical to responsible AI adoption in mental healthcare. If you want to learn more about how Lyssn assesses for bias in our own models, head to our blog. - The Psychosis Workforce Crisis Is Real
There is a shortage of licensed mental health providers in 55% of U.S. counties*, and the impact on individuals with serious mental illnesses is devastating — only 65% of adults with serious mental illnesses will receive any psychiatric care in their lifetime**, and fewer than 1% of those with a psychotic spectrum disorder will ever access CBT for Psychosis (CBTp). This is not simply a workforce shortage but a specialization crisis: less than 1% of the behavioral health workforce is trained to administer CBTp, leaving just 15 trained providers for every 10,000 individuals who need this evidence-based care. Closing this gap requires a focused, scalable investment in practitioner training for psychosis and serious mental illness interventions***. - Scaling Expert Psychosis Training Is a Major Challenge
Real-world implementation — like King County’s effort to train over 1,000 clinicians — surfaces major obstacles: maintaining feedback systems over time, supporting early-career clinicians, and meeting constant demand for training access. - Specialized AI Can Accurately Detect Provider Use of Key Clinical Skills
A study by the University of Washington and Lyssn demonstrated that AI can accurately identify whether providers are using key CBTp skills during sessions — a promising step toward scalable quality assurance. The findings emphasized that developing clinical expertise requires repeated practice paired with reliable feedback, and specialized AI has the potential to make that feedback loop affordable and widely accessible. - Early Results Show Promise in Client Outcomes
The randomized clinical trial found that clients of providers who received AI-enhanced training from Lyssn coupled with training as usual (TAU) showed significant improvements in working alliance, recovery outcomes, and reduction in paranoia when measured at 3 and 6 months — while clients of providers who received TAU alone showed no significant improvement. These findings suggest that investing in AI-enhanced provider training doesn’t just sharpen clinician skills — it meaningfully benefits the people they serve.
The partnership between Lyssn, the University of Washington, and King County’s healthcare system illustrates that the most promising path forward is collaborative — combining research, technology, and on-the-ground clinical expertise to drive impact at scale. What stood out most is how much room there is to improve how clinicians build and apply skills in real-world practice. For areas like psychosis, where access has long been a challenge, even small changes in how training is delivered could have a meaningful impact.
If scaling evidence-based practice training is a priority for your team, we’d welcome the opportunity to share how Lyssn can help support stronger skills and better patient outcomes.
References:
[1] *HRSA data, July 2024
[2] **NIMH 2018
[3] ***Kopelovich et al. 2021