 Okay, we're back here live in Las Vegas for IBM's information on demand. This is theCUBE, our flagship program. We've got the advantage to see them from the noise. I'm John Furrier, the founder of SiliconANG. I'm joined by my co-host Dave Vellante. And our next guest is Tim Buckman, professor of surgery at Emory University School of Medicine. Tim, welcome to the, to theCUBE. It's a great pleasure to be here. You guys are up on stage. You're up on stage talking with IBM. Healthcare is one of those areas where everyone's talking about big data because people can, everyone can kind of relate to that. That is one of those, those areas. First, before we get into the conversation, what, what was some of the things that you mentioned on, up on stage there? Well, the key thing is, is that we have an aging population. We have complex illness. We have chronic illnesses. My own area of Atlanta, Georgia. Those 10,000 Americans turning 65 every day. They're all migrating down. Oh, South. By part of the world. Into the South where it's warm. But what that means for those of us who provide health care is that we have to become more efficient, more effective. And then there's that little thing called the Affordable Care Act. Well, that affordable means that there's going to be less resource available to take care of each patient. So we have to find a way to take care of more patients, more complex patients, but to do it at lower cost. That means we have to change the way we do business. And as we look at our environment, our natural resource that has remained untapped virtually forever in health care are our data. We have, just now in health care, learning to use data in ways that we can make decisions for patients in real time. So what do you make of the, you know, I've mentioned the Affordable Care Act, you know, Obamacare is getting all the rage. It's in the news. Some, some bumpy, bumpy starts, but as a medical practitioner and, you know, docs aren't known for being the most aggressive in terms of adopting technology historically, but you obviously are a visionary. You see the potential. What do you see there here? A little, a little blip in the road, you know, technology, technology always getting in the way. But technology can help solve a lot of these problems too, can it? Technology is not an obstruction. Technology is an enabler. At the end of the day, the goal line is to use high tech to enable high touch. And the way I look at this is to say the current way we do business is a lot like the way my dad used to take us on Sunday drives. We would have head out of New York City and we'd hit a gas station because he'd already gotten lost. You get a map, a space of infinite possibilities. That's kind of like what our formularies are today, a space of infinite possibilities. Well, maybe I'll choose this one at this dose, okay? And then my dad would drive another 10 miles, get lost again, have to stop and ask for directions. Well, that's how we still are in medicine. Move a little bit, get some tests, move a little bit, get some tests. It sounds right, but it's probably not the most efficient way to do business. Look at what we do today. When we want to get from point A to point B, we get into our vehicle and we bring up a GPS. It might be a standalone unit or it might be an app on your iPhone, but we just plug in where we are, where we want to go. And then all the computations are done in the background, you get that nice little magenta line and it says if you follow this path, not only are you going to get where you're going, but you get some estimates and predictions around it about when you're going to get there. That's how we have to change our practice of medicine. We have to be charting paths for patients. We have to be planning our interventions and we have to be tracking the patient's trajectory. We have to be predicting where they're going to go so that we can see as they begin to veer off course. We don't want patients to get sick. We want to keep them healthy. The best way to do that is by having the analytics on the patient so we understand it's the earliest possible moment when that patient's going to veer off course. So you were describing in your keynote, you know, a patient in an ICU, they've got, you know, tubes in the nose, they've got, you know, pins. We have tubes and orifices, natural and unnatural. We have six life-sustaining drugs, each running in a separate pump. We have a machine that's pumping gases in and out of the lungs. We have a device that's cleansing the blood when kidneys fail. Each of those devices has got not only data but alarms on them. And what we have to do is create an environment so that the caregiver at the bedside can get back to the business of giving care and is not in the business of monitoring all these separate pieces of technology. We need a way to integrate those data into a coherent picture, just like that GPS image so the caregiver can focus on the journey not on the little manipulation. And having the options too, like a GPS, hey, I need some gas or you need to have some new things delivered in health care, that's maybe care, right? So the question always comes up, what about ease of use? So data science is really hot right now but like people in the trenches need an easy way to interface the data. What's your take on that? Where are we in that evolution? We're just beginning with visualizations. When you look up at the bedside monitors in an ICU or an ER, I mean, you've seen them on the TV shows with the little squiggly waveforms going across, you're getting a six second snapshot of the patient. Well, that's very useful for those six seconds. But would you tell me what the patient looked like a minute ago or 30 minutes ago or an hour ago? And if there are 20 patients, nobody can keep track. No human being can do that. But those are data in motion. And data in motion provide an opportunity for real time analytics in different types of visualizations. It's all about situation awareness. It's not enough to just perceive the data, not enough to just see the squiggly line or the number next to it. You have to comprehend it. And that usually means either trending data or understanding how two different data types are converging. And you not only need to comprehend the data, you have to be able to project into the future. I mean, when I go somewhere on a trip, about 10 days in advance, I go to one of my favorite weather sites and start looking at the forecast. I do that not because I'm interested so much in the specifics of the forecast, but I want to know if the actual weather is tracking with that prediction. So if the weather and the predictions are tracking, boy, I'm going to trust that forecast. I know exactly how I'm going to pack. But if things are not working out quite right, I'm going to put in all sorts of safety mechanisms, an extra sweater or raincoat, whatever it happens to be. Those are the types of things that we have to start doing in health care, not just doing something, see what happens, but actually making patient level predictions. So we begin to understand which patients are responding as expected. That's what personalized medicine in the high intensity environment of the ICU is going to be all about. How will this all affect training? Are you looping the data back into a training curriculum and learnings that you can leverage from the data? We have to. And in fact, we're starting to do that. I'm also a pilot. And when I learned to fly, everything was little round gauges and a lot of mental work trying to figure out what the little round gauges were telling me about where I am. When you teach folks to fly today, you teach them not only to fly the airplane, but also to manage systems. You also train not to do the routine things, but to handle the unexpected and the emergencies. In fact, we're not just going to be caregivers. We're trying to help the patient and family manage their health. Oh, it's a very different mindset as to how we approach the most important people in our work, which of course are the patient and the family. We have some commentary on our crowd chat, Tim Kudos to Tim. He says, data, situational awareness and real-time analytics in health care. Huge upside, but are we ill-prepared? Are we ill-prepared? So what's your take on that? I'll see. It's early days, right? And you've had experiences in your data practice. And what were some of those challenges? How did you get prepared? And one, are we ill-prepared today because it's early? Or what's your take on that, his comment? I think we're ill-prepared in the sense that we used to look at the delivery of care as a doctor and a nurse. Typically, doctor says nurse does. And that was the way we did health care. And however the doctor decided it was the right thing to do and however it came out, that was as good as it could get. That's kind of like the airlines were all the way up into the 70s. And although aviation safety had improved, we got to a point in the 70s where we thought we were good. And then there was a big crash in the Canary Islands because we failed to communicate data effectively. Two 747s collided and it was the largest accident in history. And we understood from that the two things had to come into play to make aviation safer. Number one, we had to get the data accurately and we had to get it in real time. We changed the way we communicated. We changed what we communicated. And second, we changed the way people interacted around data. It was no longer one person making decisions. It was teams focusing on the entire situation. That's where we're going in health care. And it's all going to be about teams of caregivers focusing on the patient and the family represented by the data. Collaboration. It is fundamental. It's all about collaboration. So you're a visionary. I can tell just by listening to you speak. What does the ICU of the future look like? The ICU of the future is a place where people come because they know they're going to get safe care. What we're going to have are data flowing out of the patient based on inputs that we put into the patient and that GPS-like picture, that representation of the patient, how they're doing in real time. But most importantly, we're going to be able to look just over the time horizon in the sense that we're going to see deviations from what's expected earlier and earlier, and intervene so fast that from the standpoint of the patient and family, we've not just mitigated the problem. We've preempted it. We've prevented it. We will be able to predict, based on the characteristics of individual patients, what the most likely problems are going to be. We'll have sensors out there detecting them, and we will intervene as soon as we see the patient getting into trouble. How do you anticipate, Tim, handling something like this? Because I can see a situation where, say, an octogenarian has maybe not as great a chance of surviving after, say, some trauma, yet if you can intervene more proactively, you might be able to predict pneumonia or something like that and keep that patient alive, even though the outcome might not be as great as you would like, if somebody who's younger. Is there a risk that you're prolonging the inevitable? I'm sure your industry talks about that dilemma. How do you handle that? Let's call it for what it is. Life is a disease with 100 percent mortality. What we're trying to do is to achieve the best outcomes that are desired by the patient family in the context of what's medically possible. The problem today is we don't have a clear picture about what's medically possible. You show me an ICU full of 20 patients. I'll give you my best estimates. I'm going to be wrong a fair percentage of the time as to how folks are going to come out. With big data, with real-time analytics, I'm going to have a much better chance being able to have a conversation with family, with patients, say, this is the most probable future. We're going to hope for the best, but let's all get together and plan for what the data are telling us. Yeah, so in that instance, the example I gave, in fact, the outcome is going to be better for the family, because there's less uncertainty, there's less fuzziness. You're actually using real data, not just maybe one doctor's opinion, which could vary from another's and different family members involved. Interesting. I'd like to take it back for a moment to something that's very fine-grained, but something that anybody who's been in a hospital has heard and has had to deal with. You go into an ICU and you hear a ding, the alarms go on constantly. In fact, for any patient in one of our ICUs, we're going to generate somewhere between 10 and 15 alarms an hour. So you have a 20-bed ICU that comes out to be 200, 300 alarms an hour. You've got alarms going off every few seconds. We know it's a problem. The regulators, the Joint Commission, they figured out that it's a problem. Now, we know that 95 percent of the alarms are nonsense alarms, but because they're simple, sentry, ignorant alarms, we can't tell the difference. One of the great opportunities that are going to be happening within the next couple of years is using data and analytics in motion to begin to combine these alarms into superalarms to help us sort out which of the ones that are telling us the patient is in trouble versus by far the majority is simply saying the machine is detecting an anomaly. What we're focused on is the patient, not the machine, and this is a way that we can use data flows in real time to help relieve caregivers of the burden of the noise, to eliminate the PTSD that we see in patients from having all these noises go off, to reduce the anxieties of the family because they can't tell the difference. This is going to change how people feel about it. Increase the productivity of the nurses, which ripples all the way down. Absolutely. Let's chill down on that, because essentially everyone talks about, oh, care, predictive analyst, and the roadmap thing that you mentioned, and the GBS is a great analogy, but there's some nuts in both operational issues. You mentioned the alarms. The business value is just that it's operational. There's efficiencies involved. Give a little more insight into some of the things that data can have an impact on operationally, whether it's the operation of the machines, which are throwing off data, as you mentioned. The personnel shifts. I was talking with the CEO of St. Luke's Medical Center that just managing the beds is an operational issue that's not sexy to talk about, but it's big recruits. It might not be sexy to talk about, but I'll tell you, every day I come into the ICU that I'm going to go back and staff in about six hours. I have 12 beds. I have a list of patients who are going to be operated on the next day, and I'm doing air traffic control. I'm trying to figure out how, in the next 12 hours, I can actually figure out how to make a bed for the patient who's coming off cardiopulmonary bypass. This happens every day. It's a huge deal. So what if, what if we could actually look at the patterns by which patients needed operations? What if we could begin to modify the scheduling of those patients to take advantage of the inevitable variations in demand? What if we could do what the airlines do and make sure that every plane goes out full, but at the same time minimize the number of people who are asked to volunteer to delay their operation? That would make much better use of our space, like building beds for free. You know, it costs somewhere between $2.5 and $3 million to build one ICU bed. So if I can just increase the efficiencies, I'm going to be saving money for my enterprise, making much better use of the resource, and, oh, incidentally, there's that little business of scheduling the personnel to staff the beds when the patient sits here. It's a perfect storm. What you just summarized is the meat and potatoes of the operation. That's the value if the caregivers are, one, happier, more efficient, have the tools and understand the roadmaps of the GPS coordinates, between point A and point B. That's a significant dollar change. So with that, the question is, what are you seeing right now relative to the value chains within the hospital that's the most in need of the big data right now? I mean, in the early days, it's going to get better and better, as we mentioned. What part of the value chain, if you will, that's right now focused on the most in terms of the big data? I'm still a caregiver, and my focus is on the patient and the family. And the truth of the matter is, is that we don't have great representations of patient's health. We have a database with some lab values here. We have x-rays over here. We have text that's begging for natural language processing. We have information on what pharmacies have been used to give patient the drugs. We have no way of bringing that together at the moment. If we can bring together all those disparate sources of information to the point of care where the caregiver is interacting with the patient, that would be a quantum leap. So there's a question from our audience that's sort of related. What does Dr. Buckman think about the collection of data both inside the hospital and collecting data outside where we might be able to observe activities or events prior to having to initiate a trip to the doctor or the ER? Will that data ultimately get into the hands of consumers? A great question. So let's talk about that transition from being in the hospital to leaving the hospital. Now imagine you've got your 85-year-old parent with a bunch of chronic diseases. They've been in the hospital for congestive heart failure, a little bit of pneumonia, get the diabetes under control, a tune-up. And it comes discharge time, and they get a list of about 15 drugs as long as the patient's arm gets electronically transmitted to the pharmacy. The patient goes out the door. So today do we know that the patient has had the prescription filled? Do we know that the patient is taking the medication as prescribed? Do we know that the patient has completed the course? No, we don't. So we're surprised when the patient gets readmitted within 30 days. Yeah, we're going to need to blend all of the outside the healthcare system activity with all the inside the healthcare system data, so we get a coherent picture of the health of the patient. I got to ask you about the answer you gave in your panel today. I think it was Bob asked why, Bob LeBlanc, why IBM? And your answers were IBM research partnering in analytics prowess, essentially streams. The research thing surprised me a little bit. Not that IBM doesn't have great research, but that someone in your position would understand the value of that. Talk about that a little bit. What about IBM research is attractive to a practice healthcare practitioner like yourself? What's attractive is vision. You know, when there's research that is totally purpose driven about the corporation, you make incremental progress. But when you think about the great research, corporate research entities of the past, and I'm thinking Bell Labs. And just the creativity that came out of that corporate investment in pure research, that's been hard to sustain in these modern times. But I think IBM has done an extraordinary job understanding that the next frontier is not going to be the little increment in today's product, but really visioning what might be done. John, that's so true, right? We see it all the time. R&D is just these little incremental improvements, make something run faster, or maybe simpler. So I got to ask. So one of the things you mentioned, you said you're doing a lot of air traffic control. You're a very high-priced worker in the system. And so, you know, managing flows may not be the best ROI in the sense of day-to-day operations, but let's talk about your data practice, because the goal is to automate it, right? Get the minds on. Right. So let's talk about what it is we want to do to make healthcare more efficient. Couple of things. First, we have to manage our stuff. That means that we have to use the most cost-effective treatment or therapy for the patient, okay? And the tendency, of course, is to use the new best, fastest patent, fanciest device so forth and so on, but understanding that the generic may be as good as the newer fancier drug. You do that a few million times. You actually start taking costs out of the system. So really getting the analytics on the effectiveness of our therapies is going to be key. Second thing that we have to understand is that healthcare really has driven the stuff cost down to the bone. So now we're dealing with staff. There are two things that are happening. First, the workforce is changing. We're moving from physicians to non-physician providers. We're moving from registered nurses to certified individuals. We have a much different blending of workforce. In order to use that effectively, we have to have ways of guiding them. One of the things that we're doing in our environment is using telemedicine, TEL-AICU, so we can put expensive people like me in command centers where we have two-way real-time audiovisual, high data density flowing, and I can look not just at the 20 patients in my ICU, but at the 120 patients across where it's all scalable. This is the power of data and real-time analytics. It enables scalability of our most costly resources. Awesome. Tim, thanks so much for coming to the CUBE. We'd love to go another hour with you. We have probably a whole conference around that. This concept is very, very important. Again, this just highlights the big data impact to lives and society. And again, the business outcome is clear. Value creation for the operations, at the same time the benefit to the end user is phenomenal. So thanks for sharing your data with the folks out there and thanks for the folks on CrowdChat. Well, the CrowdChat opened for another three hours and 22 minutes. So every guest will be on CrowdChat. We've got all your sound bites tweeted and documented. So we'll be right back with our next guest after this short break. This is the CUBE live at I-O-D, I-B-M-I-O-D is the hashtag. This I-B-M is information on demand conferences at the CUBE. We'll be right back.