Build, Lease or Buy: Behavioral Health Providers Debate Best AI Strategies

Behavioral health providers are increasingly using artificial intelligence (AI) to improve efficiency, personalize treatment pathways and produce outcomes.

How best to develop an AI-enabled technology stack remains unclear, however.

While leasing AI technology from vendors is a relatively easy path to keep up with the digital Joneses, providers with more specialized needs or those looking to increase their value proposition may be better suited to build their own tools from scratch. For behavioral health providers prioritizing speed, M&A may be the quickest route to technological advancement.

Advertisement

Before determining how to acquire AI technology, it is important to evaluate what problem needs to be solved, according to Matt Mohebbi, head of AI and research at behavioral health provider Brightside Health.

“Especially when these [large language models] were coming out, there was a little bit of an AI hammer in search of nail phenomenon going on,” Mohebbi told Behavioral Health Business. “There were definitely some cases where people would come to me saying, ‘I want to use this solution,’ and I would back up and say, ‘Well, what is the problem that we’re trying to solve? Are we sure this is a problem, and then are we sure that AI can actually solve it?”’

San Francisco-based Brightside offers therapy, psychiatry and – since its acquisition of Lionrock Health – substance use disorder (SUD) services. The virtual provider has raised more than $108 million in capital, including a $33 million Series C round in March led by venture firm S32. 

Advertisement

Build vs. lease

Brightside has leveraged AI since its infancy in the form of a clinical decision support tool called PrecisionRx. The technology makes suggestions to psychiatrists to aid in determining which medicine is the best fit for each patient. The precision tool is optional for clinicians and built into Brightside’s custom-built electronic health record (EHR).

The provider is now piloting two AI tools that would increase efficiency. One helps clinicians with note-taking, which is more popular with Brightside therapists than psychiatrists, according to Mohebbi. The other summarizes a patient’s chart to allow clinicians to quickly digest a large amount of information before an appointment, which is especially useful for ongoing encounters.

Brightside has built some of its AI capabilities in-house but leases others from vendors using a relatively simple litmus test.

“It’s looking at how unique our problem is,” Mohebbi said. “If it’s a problem that’s more unique to us or unique to mental health, then often it does make sense for us to build it in house. If it’s a problem that exists – either throughout health care or throughout technology companies – it makes more sense to outsource or to work with a vendor.”

The best-in-class AI models are typically the most expensive, but Mohebbi doesn’t start an AI initiative thinking about cost. Instead, he advises his team to start with what would be possible with the best available model. Then, if the tool does not represent a sufficient return on investment, Mohebbi will begin to cost-engineer a solution.

Along with solving specialized problems, building a bespoke AI system can also increase a behavioral health provider’s overall value, according to Dr. Sarah Adler, founder and CEO of digital mental health platform Wave. 

Adler also works as a clinical associate professor of psychiatry at Stanford University.

“It was the core value proposition for the business,” Adler said about building the company’s technology system. “Ultimately, we understand that return on investment is the thing that is going to allow us to highly differentiate, … and demonstrating outcomes is what translates into a return on investment for the people who are paying.”

Palo Alto, California-based Wave offers personalized, interactive digital tools and on-demand coaching aimed at Gen Z users. In March 2023, the provider raised $6 million in seed funding in a round led by Santé Ventures. Wave Life’s platform currently has 7,500 users and employs 22 W2 coaches.

“We wanted to make sure that we built the technological infrastructure to understand how to collect the outcomes data that we had,” Adler told BHB. “We could build on top of that, but that would be the key differentiator in our ability to sell into employers and health plans.”

Wave began as a direct-to-consumer business to validate its care model, but Adler knew that path would never create an economically solid operation. Wave Life is now pivoting to a primarily business-to-business sales motion.

The provider’s technology stack uses what Adler calls “unsexy” AI, otherwise known as predictive analytics. The software helps determine which digital tools or care pathways best suit each user.

Three smartphones show Wave's platform interface. Wave
Wave’s Recommengine tailors content to each user.

Wave’s platform is designed to treat lower-acuity behavioral health conditions. Its AI technology can personalize digital tools and suggest when users require more intensive care. 

For patients with more serious issues, Wave can refer to one of its behavioral health partners, including eating disorder provider Equip or digital psychiatry provider Talkiatry. Equip and Talkiatry, in turn, can refer its patients to Wave Life for step-down care. 

Additionally, Wave is developing generative AI tools that personalize the platform’s avatar and voiceover to reflect a user’s cultural background.

While AI is exciting, Adler said, it also requires ethical responsibility, leading Wave Life to opt against developing other generative AI-based tools like a chatbot.

“We’re using it in a very constrained way because what we found is, not only is that the best way to keep it safe, to keep it from jailbreaking [and] hallucinating, but AI right now is most popular with people when it’s not trying to traverse the uncanny valley when a human being is perfectly aware that they’re communicating with a machine,” Adler said. “We’re happy to play in that playground until we get to a point where we really can say this is going to be clinically effective, this is going to be safe.”

The M&A route

The upshot of acquiring an AI-enabled company is speed, according to two behavioral health providers that recently acquired AI-powered companies.

“There’s quite a bit of work that it takes to cultivate [AI],” Thriveworks CEO Dan Frogel previously told BHB. “From the speed to market perspective, being able to inherit a really high functioning, high-quality app makes it go a lot faster.”

“The technology now is moving very fast and in a great direction, and we picked up a lot of time by acquiring this asset,” Frogel added.

Mental health provider Thriveworks acquired Nashville, Tennessee-based Synchronous Health in early October. Synchronous has several exclusive enterprise agreements with medical systems and universities and delivers telehealth behavioral health care. In between appointments, Synchronous patients have access to the company’s AI chatbot, Karla.

Workplace mental health platform Modern Health recently acquired AI technology from Anvil Health and brought Anvil’s founder, Paul Clark, on as the company’s head of AI strategy.

San Francisco-based Modern Health provides individual and group therapy as well as self-serve digital resources. The provider has raised over $170 million and is backed by investors including Kleiner Perkins, Founders Fund, John Doerr and Y Combinator.

Modern Health operated a house-built AI stack before the acquisition that made care recommendations and matched patients with providers, along with other capabilities.

“This proprietary technology has been a key driver in our ability to connect members with the most appropriate care quickly and effectively, ultimately leading to better health outcomes,” Jesse Calderon, chief technology officer at Modern Health, told BHB in an email.

Since acquiring Anvil, the company has adopted multiple new AI tools focused on back-of-house operations.

“By integrating proven, market-ready solutions, we can accelerate our product development timeline and leverage a faster time to market enhancing our competitive edge,” Calderon said. “We have strategically decided to focus our build efforts on areas of the product that we believe are core to our business. This approach not only optimizes our resources but also ensures we stay at the forefront of technological advancements in our industry.”

The acquisition route had the added benefit of allowing Modern Health to bring on new talent and improve its technology team, Calderon said.

Looking forward, Calderon anticipates that Modern Health will leverage AI to foster personalized care, early detection and adaptive triage for patients, and to improve efficiency for clinicians.

New York City-based Resilience Lab also recently expanded its AI capabilities through acquisition. The hybrid mental health provider acquired virtual psychiatry provider Options MD, which co-founder and CEO Marc Goldberg said was an “accelerant” to achieving the company’s goals.

Resilience Lab originally built its own software stack, Goldberg said, which was a “significant investment.” The company has data from 300,000 therapy sessions accumulated using its bespoke software.

“What we [wanted was] a set of tools that can leverage that data to give the clinician insight that they will use,” Goldberg previously told BHB. “What Options MD [has] done with their AI tool is a great prototype of what we wanted to build, which is a set of data-centric tools to support the clinician.”

The new AI technology will make recommendations to clinicians based on survey and session data, resulting in more precise care, according to the company.

However it is acquired, personalized care was among the top benefits of behavioral health AI tools named by industry executives.

“AI has the potential to solve some of the problems that technology caused in health care and mental health care,” Mohebbi said. “[Now if you go] see the doctor you’re starting to remove the connection by introducing the technology. If we use AI in the right way, we can establish an even stronger connection and build better outcomes.”

Companies featured in this article:

, , , ,