Kathleen Wessel (Host): Rising costs and financial pressures, aging populations, and greater prevalence of chronic conditions—healthcare is more complex than ever before, and data is one of the keys to meeting these challenges head on.

Welcome to AHA Associates Bringing Value, a podcast from the American Hospital Association. In this series of podcasts, we speak with AHA Associate Program business Partners, check in on their efforts and learn how they support AHA Hospital and Health System members. I'm Kathleen Wessel, Vice President of Business Management and Operations at the AHA.

And today, I'm thrilled to be joined by Dr. David Feinberg, Chairman of Oracle Health and Life Sciences. In today's discussion, we'll explore how advanced analytics and cloud-enabled data integration can transform clinical and care coordination intelligence. We'll also learn valuable insights into how unified actionable data can reduce care variation and drive real improvements in population health, while supporting frontline clinicians and back office operations. Dr. David Feinberg, welcome to the podcast.

Dr. David Feinberg: Thanks so much, Kathleen. I'm excited to be here.

Host: Our audience is likely very familiar with your professional journey. You know, everything from UCLA to Geisinger, to Google to Oracle Health, which is where you are now. You've accomplished so much in your professional career. How would you characterize those highlights and, you know, what you're focused on today at Oracle Health?

Dr. David Feinberg: Well, I think my professional career really started when I had my patients that I was taking care of, like when you're done with training, I'm a child psychiatrist, and there's a couple patients that taught me so much that I still think I'm trying to get it better from them. The first was this 12-year-old boy, and he had a first-onset psychosis. His dad was a single dad from Las Vegas, a car salesman. They bring him to UCLA. And back then, this was, like, 30 years ago, 32 years ago, I thought it was really smart. I knew a lot of things like nucleus accumbens and ventricles and anticholinergic. And I was telling his dad all this stuff that I knew about probably first-onset schizophrenia. And the dad looked at me and said, "Do I have to build a room out back?" We had just had our baby girl who now is actually 33, so this is 33 years ago, and I just started to cry. Like, the trajectory of this person and this family's life had changed dramatically. And here I was hiding behind all these complicated words instead of actually talking to the dad about the most important thing, was what's going to happen to his son. Like, what does this mean for them?

And there was a second case at UCLA where a little girl wrote in her haiku poem that she wanted to commit suicide. She was a third grader. And the teacher read the poem that night, and then tells the principal the next day, who calls the mom, calls the dad. And, you know, "This is that lady. You have to know someone to get in." And so, they knew someone, so they were able to get in to see me three weeks later.

And that point, I was like, "Whoa, this is terrible. We have to make healthcare accessible." First of all, if they didn't know someone, does that mean they would never get in? And why aren't they being seen today? Why are we waiting three weeks if it were my third grade kid, like? You'd want to be seen yesterday. So, those patients and families really got me on this mission to make healthcare understandable, to make it so it's accessible for other people. Like, I don't think I've stopped doing that.

Now, I had a blast. I spent 25 years at UCLA. The last half I was the chief executive. Thrilled we took patient satisfaction from the 28th percentile to the 99th. I moved to Geisinger, and that was a fascinating move, because I wanted to learn the insurance side. There's not a lot of, back then, healthcare systems that were taking risks. And I did learn the insurance side. But what I learned much, much more importantly was at UCLA, I was worried about the next patient; Geisinger is in a love affair with its community. So, I really understood community populations at a very different level. We created the largest biobank in the world of whole exome sequencing with return results. We gave all type 2 diabetics food for their families to show that it was better than metformin. I think that's serving now millions of meals a year. And it was an amazing, amazing experience to go to rural America and learn about rural healthcare.

And I was a little hesitant. I'm a not-for-profit guy. I'm not selling ads to my patients. They convinced me to come. And I got to tell you, these were—some besides the smartest people I've ever worked with—the most mission-driven. And during COVID, that happened during my time at Google, you know, the health searches on search in YouTube became predominantly what Googlers or people were doing. We're talking, you know, 18 billion health searches a day or something like that. We had 50 billion impressions of our YouTube information page. When I left, I think it's 80 billion. And the trust of Google went up all over the world. And what was that? That was just getting good information out to people besides, you know, how to make sourdough bread. Like, "Should I get the vaccine?" and "Should I wash my fruit with Lysol?", "Am I allowed to go outside and wear a mask?" That was that same patient asking for information that I felt, "Wow, what a privilege with these Googlers to be able to do it at scale.

Kathleen, you jumped through one place that I went to. Before getting to Oracle, I went to Cerner. And I went to be the CEO of Cerner, the biggest EHR company in the world. So, we're much bigger than others. And when we're done with the VA, we're again the biggest in the United States. What that meant to me was what an opportunity, because that means we have more grandma's blood sugar than anyone else. And what a sacred position to have that. If we can understand everything that's going on with grandma, with her feet and her eyes and in her cupboard, we have this opportunity to really change healthcare.

Very early on in my time at Cerner, Oracle reached out to buy us. And that's now almost three years. And it's been amazing—and I know we're going to get into it—but it's been amazing to be at Oracle. And I would tell you of all my jobs now that I'm done with my résumé here with you, the best was writing Ritalin prescriptions for kids. Being a child psychiatrist was by far the best thing I've ever done. The second best is my current role at Oracle. And it was really because a lot of things came together when Oracle bought Cerner. I think AI's been amazing. I think my journey's helped me really understand healthcare from a variety of angles. And the time now, I think, is really to get back and fix it for those patients, to make it accessible, equitable, to make it understandable, to make it affordable, to make it so it's fun again to be a doc and a nurse. Well, at least, that's the part that's not on the résumé.

Host: Well, just, you know, the idea that all of those experiences are really contributing to what you're doing today, that makes perfect sense. And just to reflect, what a powerful set of, you know, foundational stories to start your career and really launch that passion for what you're doing. I appreciate you sharing those.

Dr. David Feinberg: Yeah. The reason I went into psychiatry was kind of a trip. I thought I'd be a pediatrician. I like kids and hanging out and my wife says if I wasn't a doctor, I'd be a camp counselor. So when I went to do adult psychiatry as a medical student in the south side of Chicago at a community mental health center, and I get there the first day, and this pretty aggressive guy who ran the community mental health center psychiatrist says to me, "What would you do if a patient came in with a rat bite? And here I'm like a third year med student. I'm like, "Treat the bite, see if there's infection," dah, dah, dah. He's like, "Wrong. You have to go in the community and kill the rats." And I was like, "Whoa." And so, he was thinking about mental health problems about how do you go upstream to prevent them? And it just changed my thinking completely. Like, that's population health. It is going into the community and killing the rats. So yeah, there've been those things, Kathleen, that you didn't expect and they kind of send you in a different direction.

Host: Yeah. Or get you to focus basically. So, healthcare, you know, we're constantly looking for innovative strategies to address entwined challenges of rising costs, aging populations, and staff burnout, among a host of other things. These aren't new challenges. But as members face increasing financial pressures and chronic disease prevalence, you know, what are some of the most impactful ways data can be used to proactively address these challenges?

Dr. David Feinberg: Yeah. If I could riff a little on it and then answer question.

Host: Yeah.

Dr. David Feinberg: So, I have this opportunity in this current role to go around the world and meet with basically customers or prospective customers. It's usually the department of health in countries all over. And they all start by going, "Healthcare is different than what you know from the United States." And I go, "Oh, well what are your issues?" They're like, "Well, our doctors are burnt out. We can't hire enough nurses. There's long weights in the ER. We have a lot of chronic conditions And it's too expensive." I'm like, "Ah, it sounds like a group of problems we may be able to address." So, it's not just the US. I think when you have humans, I think healthcare is people caring for people. It's built on trust. And then, as you build these systems, they end up actually with a lot of the same problems, regardless of where you are. More questions about revenue cycle in the US than outside the US. But other than that, it's pretty much the same.

But how can data help? So if healthcare is based on trust, it is people caring for people. And that could be a professional caregiver taking care of a patient. It could be a team of caregivers. It could be somebody doing self-care. It could be somebody at home caring for a family member. It's all care. Data didn't really matter that much in the olden days becaase there wasn't that much information. The doctor sat at the bedside at your house and hoped the fever broke. Like, you know, maybe we didn't even have antibiotics or, you know, the way you diagnose diabetes was you would actually lick the urine because at 100 or 160 or 180, when your blood sugar gets that high, it spills into your urine. Like, we didn't have all this stuff. Now, we have so much stuff and so much information that we've actually, I believe, turned away from the patient, and it became kind of glued to the terminal and the information and the coding and all that. That's been the problem. But your question is how can the data help?

I think the data can help if we refocus on what we're supposed to do and use the data as a tool and use the data as a way to make it easier to provide care, easier to sort through that information. So when you go to the doctor and you got some complicated case, the doctor doesn't have to spend hours looking through your record. We can take all that data and tee it up. And then, they might actually have a question like, "Did you have a CAT Scan?" or "When was your last colonoscopy?" They shouldn't have to touch a computer to do that. They should say, "Hey, when was your last colonoscopy?" And if you don't know, we should be listening with the patient permission and be able to pull that information in so we can get back to that person caring for another person, right?

So, I think we have this huge opportunity now that we've digitized everything to get it out of the center, bring it behind, and allow that care to happen. But bring it up with not just data, but insights at the right time for whoever you are—a patient, a caregiver, a professional caregiver.

Host: Yeah. I think having access to it when you need it, where you need it, is the key point there. So, what can Oracle Health do to help members, AHA members, in terms of defining the role of analytics, enabling better clinical and care coordination outcomes?

Dr. David Feinberg: That's our mission, is to take this information in a broad way and make it useful for your members and everyone else, right? And so, let me kind of divide it into two things. One is kind of that core electronic health record. It needed to be fixed. We've done that. I think it's always best to describe it from a patient view. You know, patient gets a text, you have an appointment Wednesday at one o'clock, say yes—press one for yes, two for no, and three to reschedule. We're like, "No, no, no, no, no. Write anything you want." "Oh, I want to talk to the doctor about metformin. I think I need my blue pill refill. Not sure I want to vaccinate my kids and my hip is hurting." We have multiple AI agents that take care of all those. Figured out what the blue pill was. It was for a different doctor, who happened to be on vacation, got to the covering doctor. Three sentences. "Are you okay with refilling the blue pill? Now, I'm seeing you on Wednesday at one o'clock," "Hey, Oracle, tell me about my next patient." And not only does it tell me that the patient has new concerns about vaccine hesitancy for the kid, and the hip problem, and the metformin is going to be taken care of, it also takes in 2000 public data sets that brings information in that could be useful for that visit.

So, I like to give the example. I'm currently living in Los Angeles. I went to the doctor here a few months ago. My doctor should have known, right, as they walk the room that I've been exposed to fire, right? Or that I come from a neighborhood where there's fume issues, right? So using not just the EHR, but multiple data sets to tee you up for the visit. Plus the little, "Hey, Oracle made the appointment 10 minutes longer because the patient had those two concerns that we learned about," so we were able to adjust it.

Now, we go in the room and like other people in the space, you know, we press a button with the patients permission, record the visit, and everything gets done. So, the doctor never, never, ever has to go to the computer. Now, they can say yes, because they wrote the note. Yes, the note's right, or make a change. Yes on the coding. Yes on the billing. Yes, I'm bringing the insurance information upfront. So, we're not going to refer you for something that's going to be pre-approval without doing the right thing. All of that stuff gets done for you.

So as we've seen this now in real life, and I've been in the room with doctors and patients when they're using this, a couple things just jump out. There is so much more eye contact in the room, like literally it is two people talk, or in one case it was a mom with her 13-year-old son for annual physical with the pediatrician. And in another case, it was a wellness check. But the people are talking to each other. It is amazing. Then, you come out of the room, and the doctors are just blown away. They're like, "Oh my God, I'm not going to retire. I get home. I had lunch this week. I've never had lunch in six years." Like, you know, yeah, the tool really works.

So, that's one piece that Oracle has done and we're excited about it. And I would say it's not just us. We see others in the field on that same journey of making it. So for AHA members, there's hope on that piece. What we do that I think is unique—actually, I don't think anyone else can do it—is combine that clinical EHR with all of those back office things: ERP, supply chain, human capital management, claims adjudication, life sciences, cloud hyperscale. So, what does that mean? Oracle's a leader—before they bought Cerner—in working with health systems on your staffing, on your budgeting, on your supply chain.

And what we're doing now—so, the second thing besides kind of fixing the EHR—is pulling that together in a platform so it's clinically informed. So, think about you hire a nurse, you train the nurse to be on the oncology unit, but you end up having to use the nurse on neuro for six months before they go to oncology. Now, they go to oncology and they're going to give a chemotherapy that's a new chemotherapy that they have not been trained on. Well, in real life, the way it works out is they ask their friend, "What do I do?" And they just go for it. We can take human capital management because we know that nurse and bring just-in-time training into the EHR. Or I'm done in the operating room, ambient or voice captures my note, but it also drives my supply chain for the OR so I don't have nurses hiding things in the operating room suites because they're worried it won't be there next week.

You know, it make sense of that end-to-end kind of platform, that I think is really, really unique to us because we offer all of those components. And then, we can double click on life sciences if you want, or on revenue cycle management or claims adjudication. But it's the same idea where Oracle has done those things, and we're bringing them in, making them clinically useful.

Host: Yeah. No, just the full ecosystem there, not just in front of the patient, but back office too. I think that that really creates just a wonderful environment where you have all of the inputs and can make all the decisions you need to because all the information's right there.

Dr. David Feinberg: And I would say, Kathleen, this differentiates us also, we know we can't do this by ourselves, so we have built a platform that is open and connected, and we want innovators to build on top of it. Let me say that again. We can't do this by ourselves. We do not think you solve problems on your own, especially as big as this one. We want to open up APIs to those EHRs, that new EHR. So if you have something that's really good, and the health system or whoever wants to use it, it will be in that workflow for that clinician.

Host: Well, that takes advantage of ideas that are coming from your staff and teams where, you know, you haven't even thought of them yet, and you can kind of bring all of those pieces together. No, that's wonderful. You started touching on some of my next questions already. But I'm wondering, you know, could you share some examples about where generative AI has or has promised to reduce additional staff burnout and enhance care outcomes?

Dr. David Feinberg: The coolest one is, you know, write the note for the doctors and the nurses. We have a nurse product too. But even the uncool ones are actually pretty cool. So, "Hey, Oracle. I'm a manager for the lives that are at risk at this hospital. Like, we have some Medicare Advantage lives, so we've taken some risk. And we're behind on our colonoscopies in that group or mammographies. "Hey, Oracle. Literally, this is what you would do. Can you set up a program for me?" And what it does using multiple AI agents is come up with what your staffing model's going to be, what you're going to need from a facility standpoint, what the ROI is going to be. So, it literally does all of the work. Again, human in the loop, you can always say no or tweak it. "Okay. Can you write the job description for the people I need for this? Well, if I don't have enough GI labs, what would you suggest I do next?" And literally an amazing time saver. And I would say, in most cases, better than human performance. All of that back office planning, staffing, supply chain, all AI-enabled.

So Oracle, our systems don't have AI bolted on top. We have AI through the whole space. I mean, Oracle's been doing autonomous database long before people were talking about AI. So, we have AI at every level of the stack integrated. When you really think about AI, it works best if it has a lot of data. So when you give it access to that whole platform, we do really well when you're saying, "Hey, compare us to somebody else who does AI for staffing or AI for clinical new documentation, because those folks got access to the EHR, but we got access to the whole thing, and we think that's very valuable."

Host: Thank you. I mean, I hadn't even thought that broadly, so I appreciate that example. That really kind of brings all of those elements to life. You know, I want to make sure I get one more question in before we have to go, but could you share with listeners kind of what trends or innovations in healthcare analytics that excite you the most? How do healthcare organizations need to think about preparing themselves for these?

Dr. David Feinberg: So, I'm going to say I'm super excited and part of me is scared. I'll tell you why.

Host: So refreshing. Thank you.

Dr. David Feinberg: Okay. Yeah. I'm actually pretty scared, because this stuff is super powerful, everything we've talked about. And it's not us, it's the whole field. When I was at Google on top of a Cerner EHR, we worked in the NHS to try to improve the time that the rapid response team could find patients that were sick, that were already hospitalized. I mean, the typical thing that your AHA hospitals deal with, like patients in the hospital. And the way they were originally doing it was if they got a creatinine level to measure your kidney function, they would then go see the patient. And it was taking them about four hours. We gave them an app on a phone, no AI, and the app on the phone got that time down to 14 minutes and decreased cardiac arrest by 30% and cost of care by 17%. Good UX, good nurses and the hospitalists, good equipment. And it just told them the creatinine went up, saved lives, and saved money. That was key.

Then, we said, "Okay, let's do AI." And this was supervised learning. This isn't even this new generative AI, because this is a few years ago. We took 70,000 patients times 600,000 variables. And we trained the computer to be able to predict who would go into kidney failure instead of just creatinine. And lo and behold, the computer could diagnose with 90% accuracy who is going to be on dialysis 48 hours before anything was apparent. So ,no lab change, no clinical change, and they'd say, "This patient"—and they were right 90% of the time—"is going to be on dialysis." So, you go from four hours to 14 minutes to negative two days. Like, you don't even know what you do as a doc. Like, this patient looks fine, but they're going to be really sick in two days. I mean, I think you'd stop some maybe nephrotoxic meds and make sure they have water to drink.

So, that's cool, right? And you're just like, "This stuff is amazing." That's analytics, that's AI at scale. Here's the part that freaks me out. That training data set was a VA data set, de-identified patients to 60,000. We then tried the application in a normal population, meaning VA is 93% male normal. Regular, it is 50% male, 50% female. And it didn't work. And so, as these models come out, how do we say, "Hey. Don't use it in this situation. Be careful in this situation. And it's okay in this situation," or even better yet, how do we get the right people and the right data in so the models will reflect the bias in the data? Like, they really get it. I don't want to be on your podcast in five years and we, as an industry are apologizing because we perpetuated bad stuff. But it's so powerful, we can't hold back. So, how do we do that and do it in a way that's very, very responsible?

Host: The comment that you made at the very, very beginning or an observation someone made to you was healthcare's different because there are consequences when things go wrong.

Dr. David Feinberg: Sure. Sure.

Host: So, you know, Dr. Feinberg, I want to thank you so much for joining me today. You've been able to share a lot about what your organization is doing and just your background, which is so fascinating to me. For our listeners, if you'd like to learn more about the AHA Associate Program or what you've heard on this podcast today, please visit us at sponsor.aha.org. This has been an Associates Bringing Value, brought to you by the American Hospital Association. Thanks for watching.