Centerstone taps tech company Lyssn to build out AI-based training for behavioral health clinicians

There is a lot of buzz about the use of artificial intelligence to help with clinical documentation and administrative tasks.

There's been less focus, however, on how AI can bring innovative approaches to clinician training.

In the behavioral health field, the capabilities of AI to analyze language opens up promising opportunities to evaluate the quality of conversations between clinicians and patients, which is a critical touchstone of effective treatment.

Centerstone, a large nonprofit mental health system that operates inpatient hospitals, outpatient clinics and other services, saw an opportunity to tap into AI and natural language processing to train its team of more than 3,000 clinicians in the latest evidence-based practices.

The behavioral health provider is working with health tech company Lyssn to use its AI-powered platform for clinician training and quality assurance programs across its wide operations. As part of the collaboration, Lyssn will also develop seven additional customized trainings for Centerstone therapists, case managers, peer support staff and other mental health clinicians.

"We want to make it easier to do hard work," Ashley Newton, chief executive officer of Centerstone’s research institute, said in an exclusive interview about the partnership. Centerstone's institute leads research, program evaluation and quality improvement initiatives across the entire organization.

Centerstone executives see the potential to leverage the power of AI to improve the way humans provide behavioral health care based on established clinical metrics, leading to better care and reduced health disparities, Newton said.

Lyssn's platform will supplement Centerstone's current training opportunities with a new avenue that allows behavioral health clinicians to practice high-quality, evidence-based protocols while the AI technology will enable immediate performance feedback, she noted.

Lyssn started as a university-based research project into how to use technology to analyze talk therapy and counseling sessions, according to David Atkins, Ph.D., Lyssn CEO and co-founder.

"So much of behavioral health care is a conversation. When we're talking about therapy or counseling or coaching, that conversation is the intervention, even thinking about medication management or care coordination, the conversation matters a lot for the treatment. The quality of the conversation in medication management is going to determine whether or not the patient follows through," Atkins said. "Because it's a conversation, it's largely a black box. We just have not had any way to actually measure and capture the quality of those conversations. That has implications everywhere in behavioral health care."

Lyssn developed AI software that has been trained to accurately evaluate evidence-based practices. Lyssn’s AI-based training modules are built on a proprietary data set that includes more than 21,000 hours of real-world sessions and 1.6 million words individually labeled based on its session analysis. Lyssn’s technology has been validated in 60-plus peer-reviewed academic publications, according to the company.

Delivering effective, data-driven and efficient clinical staff training at scale is one of the most intractable challenges facing health and human service organizations. Decades of research reveal that it often takes up to 17 years for evidence-based practices to be fully adopted by clinicians, and only about half of adult patients receive mental health treatment that is evidence-based. 

Lyssn believes AI and other innovative tech can help close that gap. Built on a decade and a half of research, Lyssn's platform uses AI and natural language processing to provide accurate, ongoing evaluation and feedback at scale, he said.

He compares it to hiring an expert evaluator to assess every conversation, every text, call or video, for the use of evidence-based practices, and then provide clinicians immediate feedback on ways to improve. 

In the behavioral health space, role-playing, expert observation and direct feedback often get the best results in training as clinicians have the opportunity to improve skills with practice.

To that end, Centerstone developed a simulation training program that includes live interactions with standardized patients to train clinicians in a variety of evidence-based practices.

"We actually started using live patient actors, and while we are getting really good feedback from learners and we're seeing the outcomes, one of the challenges continues to the scalability. When you move from a group of 30 or 40 learners to a couple thousand learners, the logistics of doing that using live actors has been really challenging. That prompted us to look at the market to see what tools and platforms were available," Newton said.

Centerstone provides 2 million services per year to 110,000 people and operates about 170 outpatient community-based clinics, residential facilities and school-based services throughout Florida, Illinois, Indiana, North Carolina and Tennessee.

"With Lyssn, we saw that as an opportunity to take a style of training that is proving to be helpful to our staff and that's helping people to be comfortably competent using evidence-based practices and do that at scale across an organization of 4,000 people," she said.

Centerstone will now have access to Lyssn's self-guided, on-demand training for motivational interviewing, cognitive behavioral therapy and multicultural counseling. Centerstone clinicians can participate in simulated clinical interactions, and Lyssn’s AI platform will evaluate responses and provide immediate tips and feedback they can then incorporate into sessions with actual clients.

Launched in 2017, Lyssn now works with 70 clinical, human services, academic and population health institutional clients. The company also works with 988 and crisis care operators to use its tech in crisis calls.

The training and quality assurance partnership with Centerstone marks the largest tech collaboration currently being deployed in the behavioral health space, according to Lyssn.

But the partnership expands well beyond a typical vendor-provider agreement, according to the executives. The two organizations will work together on long-term product development and joint marketing of new solutions for behavioral health training. The partnership also will include long-term revenue sharing on new products that are developed.

"We see this as an opportunity for us to invest in one another and the mission and the goals of our organizations. We saw Lyssn as a great partner to build out evidence-based practice training for other organizations, in addition to Centerstone," Newton said. "We have an opportunity to build out this broader curriculum that we believe is not only helpful to Centerstone but is also helpful across the industry."

Centerstone brings to the table its clinical expertise across different patient populations and clinical settings while Lyssn provides the AI and software development expertise, the executives noted.

"What we will be building are not bespoke software solutions for Centerstone. These are solutions that are broadly applicable and needed in behavioral health care," Atkins said. "We have a very ambitious timeline, and things are well underway already. There are six or seven new trainings that we'll be building together, and several of those are already in production right now."

The combination of Lyssn's technology and Centerstone's clinical expertise can help raise the bar for evaluating quality in behavioral health care, according to Atkins.

"Any website of a behavioral health provider group is going to say something about quality. When you go and ask them, 'What is it that you do to ensure high quality?' The first thing they'll say is, 'We hire licensed therapists.' That doesn't really demonstrate a commitment to quality," he said. "Centerstone shows a commitment to quality. They are actually putting time, energy and resources into having actors, people who role-play clients with clinicians, and then give them specific feedback on the skills that they demonstrate in that session."

The providers' work with Lyssn will build on those initiatives, he noted.

As a large organization, Centerstone also has a large, representative data set that Lyssn can leverage to train and build out its AI-based training modules, he noted.

"Centerstone has their fingers everywhere. They've got this large military contract, they do inpatient, they do outpatient, they do substance use, they do medication management, they do counseling and therapy. They are a perfect partner in terms of being able to develop high-quality, robust AI," Atkins said.

Behavioral health providers are exploring the use of AI and technology as demand for mental health services continues to soar. The COVID-19 pandemic "flipped the switch" on cultural conversations on mental health and helped reduce stigma around getting treatment, Atkins noted. At the same time, the fast adoption of telehealth during the pandemic highlighted that technology and innovation can help open up access to care, he noted.

Moving beyond just access to care, providers can tap into innovative technologies to focus on the quality of care as well, Atkins said. And this ties into the overall shift toward value-based care and payment models.

"One of the things that we are hearing in behavioral health care is that it's just been difficult to figure out value-based contracting when it's hard to define quality at the level of the interaction," Atkins said. "As we are out talking with payers and chief medical officers at payers, they are interested in what this AI might be able to provide in terms of some kind of quantifiable information that is a true indicator of quality, that could be involved in enhanced reimbursement, or some other basis for value-based care."

He added, "As we are thinking about how to take these products to the market, I think it sits within what's happening with the macro behavioral health ecosystem. I think that would be an interesting opportunity that we'll be looking to push as we demonstrate robustly at scale that these tools can be indicators for quality and talking to payers about that."

Leveraging tech to define high-quality care and track improved outcomes will help align different stakeholders to accelerate value-based care in behavioral health, Newton said.

"As an organization, we're thinking a lot about the role of AI, how we want to use it, how we don't want to use it, the practices that are in place, the kind of review processes that are in place," Newton said. "We've got a strong team wrapped around this, and I feel good about the ways that we're standing this up and the ways that we're exploring AI. It's an area where we want to be intentional and cautious. At the same time, we're excited about some of the opportunities it brings."