The Future of Healthcare with Larry Ellison: Oracle Health and Life Sciences Summit 2026

Hi Tom.

Hello everybody.

Well, let's see. That's really easy to say. My parents, my teachers, everyone in my family decided I was going to be a doctor when I was very young. I really had nothing to do with it. And they were, without going into a lot of detail, when I announced that I was not going to go to medical school, my parents actually stopped talking to me for a few years. They were very, very upset. I went to California, the Silicon Valley, and I said I was interested in computers. And lo and behold, that interest in computers has come full circle to become a deep interest in medicine and biology. I actually have an institute at Oxford that's really focused on genetics and life sciences, if you will, at Oxford. So, they just had to wait a little while for me to actually become active in medicine and biology. And in fact, this is why I'm still working.

If it wasn't for the progress that we're making in medicine and in biology, I think I would have retired a while ago. But in what is going to be, I think, the golden age of biology. We had the golden age of physics way back in the 30s and now it's time for the golden age of biology. Our accumulation of knowledge, better understanding, even just our understanding of DNA, it's not just the base pairs, it's not just the sequences, it's all the methylation that occurs during our lifetime, the regulation of those genes, the differential expression of those genetics, and the potential to take that understanding and dramatically improve the quality of healthcare and health span. It's just an opportunity that cannot be passed up. We are working very, very hard to build the next generation of electronic health records, but really more than electronic health records. It's really care management that we care about, if you will, to improve the quality of care. I'm sorry, Tom.

Well, one thing, the electronic health record shouldn't be a burden on people to manually fill out. We shouldn't have doctors in front of keyboards typing in and updating their electronic health records. Those should be captured automatically. And we should capture enough information in the electronic health record to meet our primary mission, which is to improve the quality of care. And we want health professionals to spend their time with patients or research, whichever is their primary objective, not to be burdened with the clerical activities of the electronic health record. That said, the electronic health record has to continue to get richer in information. As we better understand oncology, as we do more and more gene sequencing of tumors, as we better understand the mechanisms of life, all of that data needs to be captured in electronic health records so we can do a better job of caring for patients.

Well, from some of the simpler parts of AI, which is simply that AI can listen to a consultation between a physician and a patient and as the doctor changes medications, as the doctor gives certain orders, as the doctor is talking to the patient, all of those orders, all changes in medication, whatever should go into the health record should go into the health record automatically in draft form. So the doctor sees all of the things that were mentioned during the consultation with the patient, the doctor should be able to read that very quickly in a couple of minutes, okay it, or edit it, and then it automatically goes into the health record. So the AI is really recording the interaction between doctor and patient, consultations amongst doctors at a tumor board, all of that stuff that should go into the health record should go into the health record automatically. So that automation is very important. But also, if there's a new clinical trial, and again there are hospitals and hospitals, you know, if you're at Memorial Sloan Kettering or Cedars-Sinai or Cleveland Clinic, you probably have all the clinical specialties there. When you have a tumor board in the morning, there are a number of oncologists that join you in the tumor board and radiologists. Well, the radiologist should certainly be assisted when you do diagnostic work by AI. AI can see things the human eye can't. AI should assist in the diagnosis, in the interpretation of images, I think is very, very important in terms of early diagnosis. I think there's a possibility, I know we're going to get much earlier diagnosis because AI can see lesions that are much smaller than the human eye can see and pick these things up very, very quickly. So that's another, so early detection is very important. Another form of early detection, we're talking about oncology, is circulating tumor DNA. And one of the dreams is to have a blood test that will detect cancers. The problem is we can find the circulating tumor DNA but sometimes we get false positives because there's some cancer that the immune system is doing a wonderful job controlling and we don't really need to intervene when we find that kind of circulating tumor DNA. However, other circulating tumor DNA will allow us to start treatment much earlier than we otherwise could. So, it's going to be enormously helpful in diagnostics. If you look at a sonogram, I don't want to spend too much time answering the question, but if there's a knot on the umbilical cord, I'm sure a lot of you have looked at sonograms, if there's a knot on the umbilical cord, it's very difficult to see in a sonogram. And again, AI, so all of this, all of it, you know, diagnostic help, but it'll also inform the doctor if there's a new clinical trial that has just been started someplace where, but people in the tumor board in Cleveland Clinic, the team's going to know, but if you're in Wind River Reservation in Wyoming, there might not even be an oncologist there, but there might be a patient in that clinic that could effectively make use of this new clinical trial and the doctor will be immediately informed. The AI will look at the patient record and match that patient to the clinical trial if it's appropriate and the doctor will be informed. So the doctor having expert assistants, and let me emphasize with these AIs we call them agents, but another way to think of them is assistants that are providing information that are useful for the doctors, the doctors trying to come up with a care plan for a specific patient and if the information is brand new the AI will have consumed it, pass it on to the doc who then could make a decision with the patient whether that's appropriate for that patient. So there are so many examples of the AI assisting in the care of the patient.

Well, I think the patient portal that we built, we actually have a couple of patient portals. We had our original patient portal which was pretty, if you will, state-of-the-art, you know, it was one electronic health record format. It was Cerner, you know, it was built by Cerner and it was a specific Cerner hospital that the patient records were stored in a hospital system. And if it's an Epic hospital, it would be an Epic format. If it's a Cerner hospital, it would be in Cerner format. We looked at that and said actually that record doesn't belong to the hospital. That record belongs to the patient and the patient needs a full longitudinal record regardless of whether it's an Epic format, a Cerner format, an Allscripts format. So we built a patient portal that is multi-EHR. It's all of the EHRs, but also it's multi-hospital and multi-clinic. So all of your health records are delivered to the patient. So we think that's going to allow patients to better understand and if they want to participate, you know, as we're all interested in our health, the better informed the patient can be, I think the more likelihood the patient will be able to help manage their own care with the help of professionals. But you have to provide them the information regardless of what EHR the information is coming from and regardless of where that provider was, what hospital, what clinic. So we built the next generation portal which is multi-EHR and multi-hospital, multi-provider.

Well, absolutely. Well, no one's going to forget COVID for a while. And so there's two aspects. There's one, the public health aspect of all of this, being able, if you're a government official, just knowing how many empty beds you have in New York. You know, we sent a ship to New York well-intentioned. Well-intentioned because we thought we were running out of hospital rooms in New York. The fact is we didn't have a clue. I mean each hospital, now you'd have to call up all the hospitals in New York and then add them all up.

So we have, by building systems that are multi-EHR, multi-provider, you have a much better picture of how many people you vaccinated yesterday, how many rooms you have. People were worried about running out of certain kinds of equipment to treat COVID patients. We really didn't have a good grasp of inventory. Now having a multi-provider, multi-EHR, you get that New York view, New York state view, national view of the healthcare system where you know where you are. That's one thing. The other thing is the relationship between the payer and the provider. Obviously there are certain rules. I mean, not everyone can get a GLP-1 in the UK. It's an expensive drug. And there are certain rules before you're allowed, body mass index, there are a number of rules that apply to either qualify you or disqualify you from getting this particular medication. And there's this sometimes laborious negotiation between the provider who would like to get reimbursed, you know, the prior authorization, and the payer that wants to adhere to a certain set of rules set by the Congress or whoever set the rules, or the insurance company, whoever set the rules. The great thing about AI, it reads every single rule in excruciating detail. So you're much less likely to ask for prior authorization if you know you're not going to get it because the AI will tell you well your body mass index is too low or there's some other reason why you can't get this hip replacement which is again an expensive procedure. But you'll be informed early on, the provider will be informed early on what drugs are available for reimbursement, what procedures are available for reimbursement. It won't be this back and forth, back and forth which is very frustrating and expensive for both the provider, frustrating for the patient. We'll handle that much more efficiently because we simply have access to all of these rules immediately in the AI system and in the Oracle system.

So it's very important that the system keep track of, I mean managing the medical staff is enormously complex. Because, and it varies by the way, country to country. This is a chance for me to say the rules, by the way, from state to state in the United States, from state to state in Germany. Sometimes there are national standards, sometimes there are state standards. AI is perfect for this because we know what the rules and the standards are and how they vary from state to state and what credentials are required from country to country. That again will make it much easier to manage the staff, paying the staff. A lot of healthcare professionals have a private practice to go along with and they're affiliated with maybe more than one hospital, which is very often the case. We can manage all of those things because we manage paying the people. We have an HCM system. We have a supply chain and inventory system that helps you detect when your inventories are low. We can even help you find things. The interesting thing I find at hospitals, inventories are not centrally located. The hospital inventory is all over the place and if someone is bleeding and you're looking for the nearest vial of tranexamic acid to stop the bleeding, we can tell you exactly where in the hospital that's located. In fact, we could put a little map on your smartphone and get you to the right closet because people move around the inventories in hospitals. And it's a very unusual environment. It's very different than a manufacturing plant. So, we have systems to track all of that stuff and hopefully make it easier on everybody.

Well, I think all of us were a little bit shocked after COVID and improved our understanding of the state-of-the-art of medicine. And one of the things we're working on at Oxford, and I'll try to be brief, is this metagenomic blood test. We talked about circulating tumor DNA. So that's taking blood and sequencing all of the interesting, all of the DNA inside of the blood. The way we do pathology right now is actually terrifying because, maybe terrifying is too extreme, we still have petri dishes and we culture infections and it's like oh my god I was doing that in high school and that's still how we try to figure out exactly what infection this person has. I mean PCR does a lot of it but in some cases we really don't know what the fungus is or what the bacterium is or what the virus is and the only way to discover, let's say, a novel bacterium is to gene sequence that. So what we're trying to do is on a single blood test gene sequence all of the DNA in the blood. So all of the host DNA looking for circulating tumor DNA and things like that. All of the guest DNA looking for infections. We might find something like E.coli but find out by sequencing, and only by sequencing can you discover if it's resistant to antibiotics. You have to look at the entire sequence. PCR is not going to give you the answer. But we have to make that blood test, what's called a metagenomic, you know, look at all the DNA in the environment, in this blood environment, tell me everything that's there, you have to make that test fast and cost effective. If you can do that, by the way, if we had such a magic test, and by the way we're developing that test at EIT, the metagenomic blood test, if we had such a test we would have discovered a novel virus called SARS-CoV-2 far before we finally came to terms with it.