Innovative Care Solutions and Implementation with Dr. Shivdev Rao, co-founder and CEO of Abridge

[00:00:02] Hi, everyone. Welcome back to this Outcomes Rocket and Healthline Media series on the future of AI in healthcare. I'm your co-host, Dr. Jenny Yu, Chief Health Officer at Healthline Media. And here, back with Saul Marquez. Saul, it's great to be with you.

[00:00:18] Hey, it's great to be here again with you too, Jenny. Excited for today's episode. I am so excited to have Dr. Shivdev Rao, founder and CEO of Abridge, with us today. He's a practicing cardiologist and a physician who is passionate about building technologies to improve the doctor-patient relationship.

[00:00:36] I've had clinical experience with Abridge, and I've been excited to see the growth and adoption of this technology at the point of care. Shiv, so nice to have you on today. Jenny, Saul, it's great to be here. Thank you for the opportunity.

[00:00:48] I want to talk about the advancement of AI technologies currently and how it's streamlining workflows and really helping to curb physician burnout. Yeah, absolutely. I can maybe set a little bit of context on myself and the company and what we're all about.

[00:01:04] So we started Abridge in 2018, and really everything that we're building at Abridge is based on a simple idea that has had profound product implications for us. And the idea, the thesis is that healthcare is about people, and they're having conversations.

[00:01:17] That dialogue strives so many of the workflows in healthcare delivery. And they're upstream of clinical documentation, for sure. Also upstream of the orders we place, of the diagnoses we need to also code for. Upstream of so many different processes.

[00:01:33] And you and I both know when we're seeing patients immediately afterwards, top of mind for us sometimes is that we're serving multiple masters, if you will, for every single patient that we serve.

[00:01:44] And we have to write a note, for example, that other clinicians can very quickly read and grok where we're coming from, why we made specific decisions diagnostically or therapeutically. We're also having to think about revenue cycle and maybe risk adjustment if you're a PCP.

[00:02:01] And what are those coders, auditors going to think of my note when I say heart failure instead of acute on chronic systolic heart failure? And then finally, there's the third constituent. The most important person is the patient.

[00:02:15] And if they read a note, in my case, they see a term like transcatheter aortic vovuloplasty. What happens then?

[00:02:22] And I think that ability that we have, I think with this technology, with generative AI to serve multiple masters, to do that sort of style transfer from one conversation and help me focus on the person that's in front of me, the most important person, the patient is really where it's at from our perspective and what we're all about.

[00:02:41] Yeah, I love that context. And I think you're right. EHR was built for billing revenue cycle.

[00:02:47] And in while it was innovative, I don't know, 10, 15 years ago in terms of taking health care and doctors away from the mounds and mounds of paper, it then I would say hit its kind of point of it created more workload for the physicians and just how you share the story of serving different masters.

[00:03:08] I love this discussion and there's no doubt Gen AI is here to stay and it's making a huge impact. I've seen on social media, like all over the place, testimonials from physicians saying how it's changed their lives. They're having dinner now with their families.

[00:03:25] No doubt it's helping physicians. How do you see AI powered tech such as a bridge improving that patient consumer experience?

[00:03:33] Yeah, I think those testimonials that you're reading about folks having less pajama time, less time at night writing their notes in their pajamas or more time with their families having dinner. That's really scratching the surface of the value that this technology is creating.

[00:03:51] We have one clinician, 30 year career who recently told us for the first time in maybe decades, she's having eye contact with her patients and it's actually slightly uncomfortable. She's getting used to it. And I think that eye contact will translate into improved patient experience scores.

[00:04:08] And there was an article in the Journal of General Internal Medicine that was published, I think late last year, that suggested that doctors need 27 hours a day to get all of their work done. And that's not good for clinicians, but also that's not good for patients.

[00:04:23] That means that our ability to be creative in terms of designing care plans that are going to help people get where they want to go with their health is compromised.

[00:04:33] And so the ability to have that much more time in front of your patient thinking through their goals of care or what's the right medication that you should be on, I think that's where we're going to start to see the most profound impact for this technology. Love that.

[00:04:49] And I guess the thing that comes up there is bringing back the human side of the equation, right? That eye contact is priceless. Yeah, a hundred percent. Most of my dad, retired cardiologist, he retired because he couldn't type fast enough.

[00:05:03] He just felt like he was out of date with where healthcare is going. And I think one litmus test for myself personally was when I demonstrated the technology to him and he said, Hey, I should go back to practicing medicine.

[00:05:16] I think that's the moment that we're in for sure. I think in terms of being able to help people do what they went to med school or nursing school to do, which is really to build relationships with folks, with patients. Powerful.

[00:05:30] And I love your comment about building relationship. I was just going to say that the doctor patient relationship is an important one or just in general, any of the healthcare professionals and the patient. That's how you build trust.

[00:05:41] That's how you know exactly what behaviors that the patient is going to engage in, in terms of being compliant with treatment and care plans. And that relationship building, going back to the history of medicine is so important one.

[00:05:57] And so having that eye contact, having the touch, having all those things are just important in the kind of the overall story of the patient and the provider sort of experience.

[00:06:08] And when we think about those, the storytelling or the narrative in which a patient is sharing with the doctor and those conversations are being had in the set care setting. And you think about that then being data.

[00:06:20] We always think about all of the billing reimbursement data as being such widely available data that is being used in some of the AI and machine learning models. What then, how do you think about those kind of unstructured data, the conversation data as sources?

[00:06:36] And what do you see as the value now that this is the technology or bridges serving the value of giving time back, giving the relationship back. But then on the other side, what is that the value of that data?

[00:06:51] Yeah, I think there's a reason why healthcare has rushed to adopt this technology as quickly as it has like historically.

[00:06:59] So a little bit of background on me prior to this, I used to be an EVP and corporate VC at a large health system in Pennsylvania called UPMC and violated Peter's principle many times.

[00:07:08] But I got to learn osmotically from any number of different types of folks, whether it was finance oriented people or doctors or product people and certainly technologists and scientists. We funded a lot of research at Carnegie Mellon in the machine learning and AI department.

[00:07:24] And when you think about this technology, generative AI, like it's getting adopted so fast.

[00:07:29] I remember as a VC, if we were in a boardroom with a company in our portfolio and they mentioned a little bit of traction, maybe like a couple health systems in a quarter or a year, maybe even. It could be high fives all around.

[00:07:44] And I think this moment right now in time is historic and it makes sense. I think there are reasons, for example, related to data, as I think you're suggesting over 80% of the record people say is unstructured data.

[00:07:59] And we haven't been able to leverage it in ways we could, ways we might.

[00:08:04] There are any number of health plans out there who've been using for decades a type of machine learning or AI to maybe predict who's going to get readmitted to the hospital, which sickle cell patient needs care management at home.

[00:08:19] And the outputs from those models look like more structured, simple numbers, maybe like number length of stay predictors. Or they could be categories. For example, in the computer vision space, so many companies out there will predict whether a nodule is more likely to be benign versus malignant.

[00:08:37] And I think what's so exciting about this space that we're increasingly figuring out how to leverage is that the outputs can look like unstructured media objects. Now, in our case, it's a note. It's and you use the word story.

[00:08:49] And that is absolutely a first principle of clinical documentation. Going back to the 50s and Dartmouth and all those rooms where people tried to, with first principles, not with fee for service revenue models in mind, tried to think about what clinical documentation should encompass.

[00:09:04] And it's really bookended by a story. There's the patient story in the history of present illness. And then there's the assessment and plan, which is the next steps, like the adventures that we're going to choose together with our patients to help guide them where they want to go.

[00:09:18] And all of that data, all those stories are essentially unstructured assets inside of medical records that we haven't figured out how to tap in the same way that we have those structured like data that health plans have used, for example, to predict which patient's going to get readmitted.

[00:09:35] And so we have this moment now where structuring, for example, in the case of a bridge symptoms, diagnoses, procedures, medications, being able to map that data to existing data dictionaries, ontologies like you, MLS, RX norm, for example, means not only can we get to the next level of the clinical workflow, like order entry and be able to populate that inside of a CPOE and order entry system.

[00:09:58] Not only could we predict what the ICD should be, the diagnosis to help with a problem based charting workflow or diagnosis aware notes, but we can go also go to the next level and start to think about population health.

[00:10:11] How could we, for example, better understand what side effects people tend to subjectively share with their clinicians when they're on the latest JLPY? Like that's the sort of insight that we were now, I think, going to be able to use this data for as well.

[00:10:27] Yeah, and I think this is the exciting part for me, I think, to be able to tease out all of those story or narrative and being able to use that data as insights, as you mentioned, that then can really help to predict how can we go more upstream in terms of managing people's behaviors so that we can have less disease, less severity of disease that overall helps with a patient.

[00:10:51] Yeah, for sure. And then even just going home, right? I remember being with my grandfather, he several years ago, he got his pacemaker and my mom couldn't make it. So she's like, Oh my God, can you go get grandpa?

[00:11:02] And I remember getting this long list of things that I had to do. I was overwhelmed. And I just think about like a patient today, how a technology like this could really help make it so much more digestible. That's super promising. A hundred percent. Yeah. A hundred percent.

[00:11:19] Yeah. So we when we started the company, I think, and built into the name of the company as well as this idea that the most important people in health care, who are they? There's patients at the center and then there's their care teams around them.

[00:11:33] And if you can thread the needle through the patient first, but into their care teams, then there's an opportunity to create something of incredible value. Like you can start to get it value for health systems for sure.

[00:11:46] We have two out of five doctors saying they don't want to practice medicine in the next two to three years. There's a public health emergency, health systems, rural ones, especially shutting down oftentimes. And we see the amount of aggregation that's happening for financial reasons.

[00:12:01] And so can this technology assist and augment and maybe even for certain specific workflows start to automate? Absolutely. It can. We're seeing it already.

[00:12:10] For example, in our case with our use case, we're automating well over 90, sometimes 94, 95 percent of the clerical work that people have to do saving people sometimes two to three hours a day. And it's live across over 14 languages.

[00:12:24] We support over 50 under 100, but we've benchmarked about 14, 15 different languages, whether there's an interpreter in the room or not. This technology has a really high ceiling. You can't even see it yet.

[00:12:35] But the patient side of the story, I think that's the side of the story that you're bringing to the foreground here that we don't talk about as much. So patients being able to for sure today have a better experience with their clinicians because they're making eye contact.

[00:12:49] They're building the relationship. But what could we do in terms of creating a summary for the patient that will help them better understand and follow through?

[00:12:57] So live today, we are creating summaries from the same conversation, not just for the clinician, like a clinically useful, but billable note, but also a patient summary that's summarizing at the right human grade reading level what you talked about.

[00:13:09] And there's great research out of Dartmouth that patients forget all of us as human beings up to 40, 50 percent of what our clinician told us is just how we're wired. And you imagine like you're getting a new diagnosis. It's so hard to keep track.

[00:13:22] So that ability to go back and not just better understand and remember what actually happened, but also hopefully better follow through in the way that we can integrate with systems and nudge patients and help them not just improve their health literacy, but hopefully get a better understanding of that medication and why they should take it or that referral and why they should keep that appointment.

[00:13:44] That's amazing. And just like a side question, this is not even in the list, but OK, I'm here, right? Two amazing doctors with me and we're all here together. Would you guys, you guys are doing great work.

[00:13:55] You left the bedside, but would you still be, would you still be a physician? Do you think if something like a bridge existed before you made that decision? And this is for both of you guys. Yeah, that's a great question. For me personally, I'm still seeing patients.

[00:14:07] And I'm still practicing. So one weekend a month, I'll see patients and I'll see if nobody else wants the department and I get to stay up to speed. Fellows teaching me basically, and I'm reading some echoes, doing some studies and selfishly also using a bridge.

[00:14:23] And I mean, I think like a big aspect or ethos in the company that we take a lot of pride in is as much as we've got computer science at the center, we publish peer reviewed papers and our chief technology science officer, Zach Lipton, and he's on these steering committees for AI evaluation for the industry at large.

[00:14:39] As much as we've got that level of rigor, we have a four clinicians by clinicians ethos in the company with patients at the center.

[00:14:46] But I think being able to dog food, what we're building, at least at this juncture of my life and career, I feel is an incredible privilege. So I'll take calls every Thursday night as well. So tonight I'll be on call for cardiology.

[00:15:00] So I'd say would I practice more? That's a good question. Maybe. But I think that all of us, maybe Jenny, you can relate to this at some point in our career, like our medical careers, we realized that impact one life in a week.

[00:15:15] And there's a sense of fulfillment that you get that is bar none. It's like hard to compare.

[00:15:19] And surgeons doing a surgery and whatever your PCP, just making a difference in helping a patient or a family member stay on track with whatever their care plan is that makes the most sense for them.

[00:15:32] But I think some of us also get really addicted or enamored by this idea of scale, impacted scale. And it's not that one's better than the other, but if you're trying to go at impacted scale, you have limited options. You can do clinical research. That's its own treadmill.

[00:15:48] You can do basic science research. You got to have a really high attention span, like 13 years bench to bedside. You got to get comfortable with my experience. But then there's some of us who recognize that tech is a way forward to get at that level of impact.

[00:16:02] And I think with humility, I can say that it's not better than the impact that any given PCP is creating on a daily basis for the patients that they're seeing. It's just different.

[00:16:14] And I think Jenny, maybe you can relate to this, but I think that this like right now I'm trying to have it both ways a little bit, but I think the scale impact certainly is what gets me most excited in the morning. Yeah.

[00:16:24] And I would agree with that for sure Shiv. And I think I'm not practicing as you are, but I get to do so that patient clinical work when I travel abroad globally. And it's good to stay grounded, but like you said, the impact of scale is so important.

[00:16:39] And I just remember while I was practicing seeing all the kind of inefficiency or redundancy that didn't really have any logic to it.

[00:16:49] And I knew that there were sort of ways in which you can use technology for, but it required clinicians and scientists and different sort of groups of people with different perspectives to really think through the workflow within the healthcare system.

[00:17:05] And so whether it's surgical innovation and diagnostics and imaging, the workflow technology, whatever it may be, there's just so many ways in which technology can be helpful in healthcare. And of course it's, of course it's, it needs to have validation. It needs to have the ability to scale.

[00:17:23] And I would say Sol that your question of, I certainly own up to the fact that I was someone who Friday night spent time with my Epic charts trying to close it and drop the note so that I don't forget what I has said to the patient.

[00:17:37] But I always valued my time with the patient. So I, I suffered as a result of the not actually documenting behind the computer screen while I was with the patient because I felt that was important.

[00:17:49] And I just felt like that's not scalable and that's not sustainable from my time away from work. And I agree with Shiv's comment about the, our addiction or enamorment of the impact of scale. I appreciate it both. And that was totally an unplanned question.

[00:18:04] So thank you both for answering that. We have one more question. Jenny, why don't you go ahead and take yours and then we'll wrap up here.

[00:18:10] Yeah, I think at the end of the day, it's really just about what does this mean as you see this shift as its growth and adoption, both from a system perspective in terms of the operating workflows, and then also just from a privacy sense.

[00:18:24] And so if you can close us out with that question. Yeah. So previously you'd use the word trust. And I think that technologies that can help bring people together, help them build stronger relationships are going to be the most profound in healthcare delivery.

[00:18:42] As long as you believe in this thesis that it's about people. And we certainly don't think that's going to change in the next 10 years. So this technology certainly can bring people closer together. It can make healthcare delivery feel even more human. Certainly.

[00:18:57] And I think over time at scale, what we'll start to see is it's improving not just experiences, but we'll see in a rigorous, in a data oriented way that it's improving outcomes.

[00:19:07] Like that's always, of course, the Holy Grail for any technology in healthcare that you can get at that story, that ambition over time.

[00:19:15] And what's wild about this moment right now that does feel historic is it seems as if we're going to have that kind of data before we know it in a matter of months, not years.

[00:19:24] From a qualitative perspective, just thinking about what this impact really means for us as we're scaling across any number of health systems. We recently announced a partnership with Sutter and before that Yale and UC Irvine and the list goes on.

[00:19:38] But this is, we have a, I'm opening up our Slack. So we have a, we use Slack as a company to communicate with each other.

[00:19:45] And we have this one channel called Love Stories where on a daily basis, we're getting qualitative feedback from our users and everybody reads every single one of them because every single one is just inspiring. In some way or another.

[00:19:55] I remember when we started the company, we used to joke about learning to digest glass as an early stage startup. You just have to figure it out.

[00:20:03] And we probably all saw that video of Jensen doing a fireside chat and him mentioning that it's resilience, it's character that actually is the strongest predictor of like a successful company.

[00:20:16] And that comes from being able to take punches, but just always come back, get back up and figure it out. And another metaphor that we've used too, is like being a hope camel, like a drop of hope should get you through a desert.

[00:20:29] So every single one of these entries in our Slack is a drop of hope and now it's just like trickling all day long. So just to read the last couple, here's one for example, and this is qualitative and of one.

[00:20:39] And, but you think about the impact at scale, sorry, I'm just opening up Slack. But what makes a bridge truly special is its ability to allow patients to interact with clinicians in ways they feel comfortable.

[00:20:50] With one clinician conducted a visit in both Vietnamese and English wearing an N95 mask and a complete note was produced in real time. So cool multilingual, but here's another one that may be even more affecting. This is from a PCP.

[00:21:04] I was sitting at dinner last week and my son asked me, mommy, why aren't you working right now? I literally took my phone out and explained to him that a bridge is a new tool that lets mommy come home early and eat dinner with her family.

[00:21:15] I started to tear up and looked over at my husband who then said, mommy's going to be eating dinner with us every night now. So it's that sort of feedback.

[00:21:23] And I think what's so exciting about this moment in time is that we are still in relatively early innings with this technology. We are still scratching the surface of what's possible. We have an incredible privilege of being in the workshop program with Epic, of being a co-development partner.

[00:21:40] And our folks are together practically every week, it would seem. And I was in Verona earlier this week. And I think the roadmap of what this technology can create when it's really deeply integrated inside the workflow is just like sky is the limit.

[00:21:54] And when we share what that roadmap looks like with our partners in health systems, it's like people's eyes just get so wide. I'm like, really? Is that possible?

[00:22:04] And that everybody's coming together to make this stuff possible and real as quickly as we're doing it is what's also so different about this moment right now.

[00:22:13] I think that getting back to like why we do this in the first place, I think it's like that scale at impact.

[00:22:18] You never know how long it's going to take, but that it's taking as little time as it is right now with such a bleeding edge technology is such an exciting moment for all of us. I love that, which is what you shared.

[00:22:32] And obviously knowing that all of the hard work and the blood and sweat that you poured into this company and now seeing this moment and seeing the kind of adoption of just Gen AI in general and people willing to write there's willingness to take this on, I think will be that silver lining that is going to save healthcare in such a way that we could reimagine getting it back to the core, which is the

[00:22:59] Yeah, I almost cried there on that one, to be honest. Such a beautiful story and there's so many of them. Couldn't agree with you more, Jenny. Yeah. Chip, what a great opportunity to connect with you here. The difference you guys are making really appreciate the time.

[00:23:12] Yeah, it's been a privilege. Really appreciate yours as well. Thank you. We want to thank our listeners for joining us on this episode of the future in AI and healthcare series, and be sure to check out the show notes from today.