Hi, everyone. I'm Erica Carbajal with Becker's Healthcare, and thank you so much for tuning in to this episode of the Becker's Healthcare Podcast, where we're joined by Annie Flanagan, CEO of Iris Health. Andy, thanks so much for being on today. How are you? Thank you, Erica. Great to be here. Yeah. I'm excited for this this discussion. We're gonna be chatting about how health systems can use AI to overcome common challenges when it comes to scaling behavioral health services, A huge challenge and focus right now at health systems across the country and how they can really achieve long term success through innovation. So, Andy, with that, I'll just start kind of to help lay the groundwork for the conversation. Can you recap some of the main challenges that health systems are facing today when it comes to delivering scalable, sustainable behavioral health services? Yeah. We're playing defense in our country right now. At the health system level, the cross continuum impact of improper ED utilization hospitalization is crushing us. And, unfortunately, the patient journey and the physician journey suffers in equal amounts. Really this transition of care we've always struggled with as people enter and leave acute care settings and behavioral health because so much is delivered in an outpatient manner and at the primary care level that when escalations occurred, really difficult to deescalate, stabilize, and return process as people back to a, an appropriate lower setting of care. So, you know, really, it's it's a pervasive problem. And I think there's some great solutions in the market. Certain systems are doing amazing work. But it's early. And and we all need to be watching, learning, listening from, some of the civil works being done around the country. Yeah. Absolutely. I think that that utilization piece at the start that you mentioned there on ED is just something we hear about so often, you know, from hospital leaders just around, you know, when patients are showing up in the ED and behavioral health crisis, just really a sign that the the system has failed them already. So, Andy, moving into more sort of a solutions component here, where does AI and technology fit into an innovative approach to care that solves some of these these long standing challenges and be behavioral health access? Yeah. You know, most of AI today captured the headline, is captured trying to save the physician time, ambient, learning, and in in some cases, dabbling in diagnosis. In behavioral health, where most most of the data is unstructured, and biometrics really don't play a huge factor in the diagnosis or the treatment, except, you know, when you think about medication management on a quarterly basis. You know, we really have to think about AI as machine learning. We have to think about, the RPA work that's being done already as an example of the work that can be done. These tools usually sit in a, higher reimbursement area, like oncology or cardiology. And in behavioral health, there definitely are some opportunities. You know, we ourselves have released a, an ML risk model to predict ED utilization in our primary care panel. And there are many companies that are doing the same thing. Many hospitals that are funding innovation, co pilots, things like that. But it is very specifically about thinking about, leveraging all of the data we have around us, in behavioral health that traditionally is unstructured. AI can help us do that faster. Sure. Andy, building off that, what are some of the specific ways that the AI can be deployed to improve care optimization? A great example would be as simple as, a Sentient, approach or Sentinel approach to monitor patient panels for the patients that already have an existing behavioral health indication. We often are asking our primary care clinicians to see, you know, patients every 10 minutes, 12 minutes, sometimes 9. And when they walk in and and the the prescheduling pre charting activity in the morning, how often are we presenting them with some kind of a an assessment of my vulnerability to a b h concern or the presence of a previous indication unless it's fairly recent. And so, the lack of access for therapy, the lack of access for behavioral health resources has pushed a lot of this work on the primary care. And halfway through a 10 minute encounter, 12 minute encounter, they realize they have a challenge in front of them. And this happens, you know, maybe 20% of their encounters fall into this category. So it's it's our position and our belief, and and we're investing, to help the health system address behavioral health in a distributed manner, integrating into the primary care, complementary to things like coordination of care, COCM, PHI. We're really talking about technology here. This technology can be deployed in a very simple manner with very little low clinical risk and, in a very cost effective manner. And we're talking about identifying me, a person, before I escalate. And the data suggests that that's fairly high quality work. That's a that's a reasonable expectation, a reasonable prediction. And so I believe that if we're going to focus on defense in the ED, which we are, by by nature, we need our EDs to be unclogged with the behavioral health patients that are being boarded and go on offense and start meeting people where they are and supporting our primary care network to effectively, you know, manage utilization. Yeah. Absolutely. I think that that stat or estimation there you shared around, like, 20% of the time just Number. Yeah. For primary care physicians not having that visibility when the when the patient comes in for their visit is just speaks to the the huge opportunity to to improve there. Andy, what are some considerations health systems should keep in mind when incorporating AI into their services and workflows? You know, a lot of the options for health systems, are presented by startups. There they've raised a round or a b round. They're working on a very specific problem. Most of the large legacy companies that they're working with have significant investments being made in AI right now. We're a good example. And it's easy to identify, a company that does one thing to solve one problem. I I see and I hear. I speak to many CIOs, chief digital officers, chief innovation officers, you know, on a on a weekly and a monthly basis. And they're turning to their large partners first to ask the question. Partners aren't always talking about it because it's early, and it's a small part of their business. And there are big problems going on that everyone is working on as as we seek. So my my feedback for the health system is thoroughly vet all of your existing partners first. Push them to help solve the incremental problem. And I also think that the clinical diagnosis is not the problem in behavioral health that we should be focusing on. That's not the place to start. We have a we have a workflow problem. We have an identification of classification problem. We don't have a diagnosis problem. That's my opinion. And when we get a lot better, we will. I mean, we have one. But to your specific question, where do you start? Right? Start with the workflow optimization. You can capture real time benefits. You can start at one facility. That's that leads to 1 ED. You can deal with 1 patient cohort and produce very tangible outcomes, without really worrying about, is it safe to allow an AI algorithm to diagnose me in any way? That's my advice. Sure. Thanks, Andy. To wrap us up, can you share how can systems that are thinking about, you know, how to really start embracing innovation? Any thoughts on how they can really set themselves up for long term success here? You know, behavioral health is many cases thought about as a separate service line. And our belief firmly is that behavioral health should always be on. We all have behavioral health. We all have mental health, just like we have physical health. Good or bad, we have it. And embedding innovation and and technology around behavioral health into your other service lines, whether it's oncology or cardiology, you can imagine that if I were to receive an adverse diagnosis, you know, from from an oncologist, I might need some therapy. That might be a problem for me. And and that's a logical human condition. That's a natural reaction to what we see as a digital footprint when we see an MRI scan. Why am I getting an MRI? You follow that thread in a very simple way. You can very quickly realize that I might, a, be a high utilizer in the future, and, b, I might need some more support to help me from escalating. And so I I would suggest that thinking about behavioral health as a fully integrated component of your major service lines and your primary care network and investing in technology and workflow innovation in those settings with behavioral health as part of the solution, not the sole solution. I think that allows you to have a cost effective innovation strategy in behavioral health. Andy, thanks so much. I think that sounds that sums up the conversation so well. Thank you so much for being on the podcast today. It was a pleasure to speak with you on this topic. Thanks, Erica. And I I I'd like to thank everybody listening here for what you do to help solve this problem. This we need a 1000 more companies like ours solving this problem. And, I hope everybody leads in. Absolutely. And more in the art. Thank you, Erica. Well, thanks, everyone. I also wanna thank our podcast sponsor for this episode, Iris Telehealth. And listeners, you can tune in to additional episodes of the podcast by visiting beckershospitalreview.com/podcast. Thank you all so much.