AI’s Role in Behavioral Health – reality check
Say "virtual first care" one more time.
In recent years, mental health has become a growing concern, both in public discourse and within the healthcare sector. Despite increased awareness, the statistics are sobering. Nearly 20% of U.S. adults reported experiencing a mental illness in 2022, yet over half of these individuals did not receive treatment. Among younger demographics, the picture is just as dire: around 15% of American youth experienced at least one major depressive episode, with 60% not receiving care. These alarming figures reveal the vast unmet need in behavioral health—a need that digital health solutions, particularly AI, are poised to address.
Consequently, It's no surprise that between 2018 and Q1–Q3 2023, behavioral health ranked as the top clinical area for venture funding, according to Rock Health. In 2021 alone more than $5Bn were invested in this field. But can AI driven digital health innovations truly deliver the relief that the behavioral health sector so desperately needs? What are the challenges to introduce such tools to providers and payers? Where do we see concrete progress and what is still lagging?
The following piece is a based on a recent discussion we held with a diverse group of experienced leaders in the field:
Ian Chiang, Partner, Flare Capital Partners
Katherine Hobbs, CEO, Author Health and formerly CEO of Optum Behavioral Health
Linda Rosenberg, Former President & CEO, National Council for Mental Wellbeing
Dror Zaide, Co-Founder & GM, Eleos Health
Want to stay updated on new posts ? (No more than once a month!)
The Challenges: Diagnosis, Treatment, and Access
The central question becomes: what is missing in our current approach to behavioral health? Is the issue poor quality of care? A lack of therapists? Delayed diagnosis?
Sadly, the answer seems to be all the above.
The behavioral health landscape suffers from a mismatch between supply and demand. More therapists are entering the field, but the number of individuals in need of care far outstrips the available professionals. Furthermore, those who need the most immediate or intensive attention are not being surfaced efficiently, while those with milder conditions that can benefit from lower acuity types of intervention (even without a therapists) are shuffled into systems ill-equipped to address their needs. This results in a fragmented approach to care, where people are treated for individual symptoms rather than as whole beings with interrelated conditions—be it anxiety and depression, or mental illness and chronic conditions like diabetes or cancer.
What’s the solution? For one, there’s an urgent need to streamline diagnosis and treatment. Identifying individuals earlier in their mental health journey and matching them with appropriate care is critical, but doing so requires tools that can sift through large amounts of data, triage patients effectively, and allocate resources wisely. This is where AI has the potential to revolutionize behavioral health.
What Do Stakeholders Care About?
Before diving into the potential of AI, it's crucial to understand the perspectives of key stakeholders in the behavioral health ecosystem—namely, healthcare providers, health plans, and investors. While improving patient outcomes is a noble mission, there is also a clear financial imperative driving these parties that cannot be overlooked.
For providers and health plans, any new solution must offer a measurable financial impact, whether that’s through clear reimbursement models, increased revenue generation, or cost reduction. Second, ease of deployment is paramount—solutions must integrate seamlessly into existing systems and workflows. Finally, solutions must target one of the top three priorities within an organization. Behavioral health interventions are valuable, but they must also be pragmatic and deliver quantifiable results.
As for investors, beyond a validated clinical model which is essential given the plethora of solutions that are overwhelming the market, differentiation is also critical. Given the growing number of attempts to innovate in BH, Investors want to know why a particular company stands out—whether it’s through access to unique data sets, deep domain expertise, or a demonstrated ability to delight customers.
AI’s Promise in Behavioral Health
Now, let's turn to the question at hand: what role can AI play in addressing the challenges facing behavioral health?
There are two primary avenues where AI could make an impact. The first is on the operational side—using AI to streamline workflows, improve efficiency, and enhance productivity. The second, more ambitious, goal is to use AI to transform the clinical side of behavioral health by enabling earlier diagnosis, more personalized treatments, and continuous care.
Operational Efficiency: A Low-Hanging Fruit
In the near term, like other areas of healthcare, AI’s greatest promise lies in improving operational efficiency. Many behavioral health providers are struggling with administrative burdens, from managing appointments and coordinating care to ensuring compliance with various regulations. AI-powered workflow automation could reduce much of this strain, allowing providers to focus more on patient care and less on paperwork. As an example, Eleos Health (Arkin’s portfolio company) has gained meaningful traction with behavioral health providers having the clear mission statement of “More Care, Less Ops”
But the potential goes beyond simple task automation. AI can help optimize resource allocation, ensuring that therapists are matched with patients based on need and expertise. Additionally, AI tools can streamline communication between providers, allowing for a more cohesive and coordinated approach to treatment. There are different companies who are making meaningful strides on that end such as Headway, Grow Therapy and Alma.
The operational application of AI offers a relatively quick and tangible return on investment. Improving internal productivity and reducing administrative overhead translates into immediate cost savings, making this an appealing area for early AI adoption.
The team at Flare authored a terrific extensive analysis that can provide much more color
Clinical Impact: The Long Road Ahead
While operational applications might be the more low hanging fruit, fixing the supply demand mismatch in behavioral health will only happen if we can also transform the clinical side of care. Imagine a future where AI helps clinicians diagnose conditions more accurately and earlier than ever before, or where patients receive AI-driven, personalized treatment plans tailored to their unique mental health needs.
Despite the promise, significant hurdles remain.
There are many areas in behavioral health where there are no objective markers to determine efficacy of new treatment (for example like reducing HbA1C for diabetes) which leads to a significant challenge in convincing clinicians that a new tool is indeed effective. Furthermore, the lack of clear reimbursement models for AI-based clinical tools, combined with the general skepticism about digital therapeutics (more color in this piece), has led to slow progress in this space.
Yet, it doesn’t seem like a question of “if” but of “when”, given the huge need. The eventual integration of AI into clinical care is inevitable, especially as reimbursement models evolve and AI tools become more adept at navigating the complexities of mental health. Software can and should play a role in delivering better care—it’s only a matter of time before these innovations become part of standard clinical workflows.
Overcoming the Ethical and Technical Challenges
Developing AI solutions for behavioral health comes with its own unique set of challenges, both ethical and technical. Behavioral health is an area fraught with potential for bias, particularly when it comes to diverse populations with varying mental health needs. Different people express their mental issues in different ways (based on their environment, culture, etc.) and AI algorithms must be designed with fairness and inclusivity in mind to ensure they don’t perpetuate existing disparities in care.
Moreover, as AI takes on more significant roles in diagnosing and treating mental health conditions, ethical frameworks must be in place to ensure responsible use. Who is accountable if an AI-driven diagnosis leads to an incorrect treatment plan? What measures are in place to protect patient privacy when using AI to analyze such sensitive health data? These are just some of the questions that need to be addressed as AI continues to gain traction in this field.
Looking Ahead: The Future of AI in Behavioral Health
The future of AI in behavioral health is bright, but there’s still much work to be done. In the short term, we can expect AI to enhance operational efficiency and ease the burden on overworked providers. We already starting to see this in play. Over the longer term, as reimbursement models evolve and trust in digital therapeutics grows, AI will likely play a transformative role in clinical care.
Ultimately, the integration of AI into behavioral health will depend on the ability of these technologies to deliver real, measurable results—both in terms of patient outcomes and financial ROI. If AI can prove its worth on both fronts, it could be the key to addressing the growing crisis in mental health care.
Thank you for reading "Differential Diagnosis"!
If you are enjoying it, it would be terrific and super helpful if you can please spread the word with others!