 From Las Vegas, it's theCUBE. Cover EMC World 2016, brought to you by EMC. Now, here are your hosts, Stu Miniman and Brian Gracely. Welcome back to theCUBE, live in Las Vegas at EMC World 2016. Always happy when we can sit down, talk to some of the practitioners, understanding new technologies, changing the way their business work. Happy to have on the program for the first time. I see Siva Sagaran, who is the Corporate Director of IS Infrastructure at Penn Medicine. Thank you so much for joining us. Thank you for having me. All right, so first time at EMC World, first time on theCUBE, give us a little bit about your role and what do you do at Penn Medicine? Absolutely, I'm the Corporate Director of IT Infrastructure, so I take care of operations, IT operations, as well as engineering, and I manage the architecture group, so essentially setting the infrastructure standards for pretty much all the entities at Penn Medicine. All right, can you scope out a little bit for us? You know, how large the organization is, how many locations, what kind of responsibilities you have? Absolutely, we're about approximately 32,000 employees now, as of the last acquisition, so we are five large hospitals, three rehab hospitals, seven regional medical centers, approximately 400 clinical practices, and five major hospitals. We have about 5.2 million patient visits a year, roughly about six billion in revenues and growing. Okay, so of course the challenge in healthcare is there's certain governance and compliance things you have to, but there's vast amounts of things you can do with data. Can you just give us, what's the philosophy, how do you guys think of data at Penn Medicine? Yeah, I mean it's a dual edged sword, right? You're at one point collecting a lot of data and starting to manage it and consolidate it, but at the same time it actually ends up being a benefit for us with some of the work we've been doing. The more data we collect, the more information we're able to gather, and we have some very good use case scenarios where we're able to use that to actually affect patient care in a very positive way. Yeah, I would sort of, what you're talking about in terms of collecting data, managing data, if people might call it big data, data lake, how do you think about that as a problem? You know, I doubt you wake up in the morning and go I have a big data problem, you think about it as a patient care problem or an insight, what keeps you up at night? How do you think about that? Yeah, I mean, all companies across all industries collect data, I think that's very easy to do. Some have gotten good at managing that data and consolidating it and putting it in nice neat piles, but I think there are very few that actually are able to use that data to affect their business operations on a day-to-day basis. What keeps me at night is not so much the collection part but enabling the business so we could actually make use of that data, and we always, as a health system, we treat the most acute cases. We're a highest ranked academic medical center in the country, but we're also top 10 health system in the country, and we treat some very, very advanced cases of cancer and some advanced cardiology procedures, organ transplant. So if I could somehow build a solution that could enable our patient care providers to create a solution that will make that patient feel better, that's good. And that's what keeps me up at night is thinking about how to build those solutions. So obviously, you're highly ranked, the doctors you have are world-class, they've got an immense amount of knowledge. How do you bring data-centric solutions to help them? Can you give us some examples of how you're helping patients, augmenting the doctors? Yeah, absolutely. So we have a very, very talented data analytics team and a very talented data science team. And the use case scenario is that these folks are envisioning and some are already in practice and many, many examples, but some of them are as fundamental, such as being able to do predictive medicine, which is unheard of a revolution or even like a few years ago, but it's actually practiced right now. So if you're able to, for example, identify at-risk congestive heart failure patients and determine if a heart transplant is better for them, now we have an opportunity to save that person's life. If you could wean somebody off of ventilators, on average, I heard that hospitals tend to keep patients on ventilators for 2,000 extra hours when they don't need to be on there. So by weaning the patient off, not only is there a financial cost savings, but the patient care, the patient's experience itself has improved dramatically, right? So using RTLS data, for example, to track clinical equipment such as intrusion pumps and mobile EKG machines and identifying when we have millions and millions of square feet, identifying where these equipment are located, but at the same time where they're needed, where the patients that need them are, then we're able to divert that precious resources to the location where it's actually needed. So these are some of many, many examples that we're looking at being able to predict cardiac failure and I'm sort of organ failure and being able to, I'm sorry, I think you have a question. No, no, no, it sounds, I mean, you've got a big data challenge, it also sounds like you've got an internet of things, you know, inside the hospital, so. Absolutely, I mean, we're definitely creating more data because of the emergent acquisition than I think because of the clinical workflows where each application is generating a lot of data, but IoT, I mean, smart devices, everything from a bio-freezer holding tissue samples that's reporting back temperature fluctuations to clinical devices, like I mentioned using RTLS, at locating, homing back a beacon saying here is where I am, come and get me, there's a lot of being generated so this is a never ending issue, I think data is going to constantly get generated. Boy, so I mean, you've got all the centers coming in, you mentioned that there was a recent merger that you've done, what does that do to kind of the infrastructure that you have, what do you look for and kind of the infrastructure underneath that? Yeah, absolutely, just to step back a bit, unlike other industries, healthcare IT tends to be very low in terms of percentage compared to operating revenue, right? Operating margin and revenue, so what that means is we don't get a lot of do-overs, so when we scale up and when we grow so much, we have to pick solutions that are, that doesn't necessarily require a whole platform shift or platform change or migration, so we had to pick solutions that are highly scalable and some of the concepts we're learning here in terms of making sure they're cloud-first and being able to take advantage of resources, highly scalable resources in the cloud, that's essentially the way I think we'll go as well because building on-prem and building, when you build to a certain definition, by the time you're done with an M&A, it's already out of scope and out of scale and then you had to do that all over again, which is not a luxury we have in terms of financial resources but also time as well because we were so focused on taking care of the patient care mission that this is the last thing we want to do is to worry about the bits and the bytes, you know. So scale gets thrown around a lot, do you have any metrics you can tell us kind of where you are today and what kind of growth you're seeing? Yeah, absolutely, I mean, if I were to consolidate all the clinical data, clinical and administrative data, I would say it's probably in the range of four to five terabyte, I'm sorry, petabytes, especially with some of the gene sequencing data that we have, but I could easily see that growing to 30, 40, 50 petabytes in a matter of four or five years. Data generation's just skyrocketed and that's what I mean. We don't have the luxury of putting in a footprint and assuming that's what it's going to be four or five years from now. Yeah, you talked about, you've got an outstanding data analytics team. Can you talk broader about what your team looks like in terms of skills but also, what do you look for the vendors and partners here to help augment technology-wise to help you be successful? So the data scientists are probably the smartest folks in the organization, a lot of MD, PhDs, a lot of PhDs and statistics. These are folks that are very, very good at taking information of large amounts of raw, complex information and being able to see patterns. So they're not necessarily trained to figure out what the best infrastructure solution is. For example, they're basically saying, we just want you guys to provide us this. We really don't care what the solution is behind that. It just needs to be scalable because we're going to be diverting a lot of data towards that infrastructure. Yeah, how would you characterize the relationship kind of between the business side of the house and IT? I think it's very, very good. So it's very common. I mean, I used to work in financial services where IT was sort of viewed as a cost center, right? You're there to support the mission but you're not necessarily part of the value chain whereas Penn Medicine, especially ITs at the table, which is a very unique relationship and we're looked on as a valuable member of that value chain. I always tell my team is, yes, we do IT but our primary job is to support the patient care mission and by sort of changing that paradigm, changing that framework and way of thinking, we're able to now contribute to the patient care mission rather than just talking about, I'm going to stand up the next door to Ray, which is what the discussion was about 10 years ago. Right, healthcare is obviously a unique industry. You've got your regulations like HIPAA and privacy and how do you sort of manage, how do you think about it and manage the value that comes out of data when it's all collected together but also maintaining privacy and compliance? It's a delicate balance, right? So one hand, you want to enable the business and the other hand, you have to worry about how much to hold back and how much to lock down and secure. So it is a delicate balance. I don't think there's an ideal answer. It really depends on the industry and depends on the customer themselves or the organization themselves. For us, I think we've kind of struck a very good balance where our providers are able to tell us what they need and how much access they need and with good governance, with good administrative procedures, with good process and good guidance, I think we're able to strike that balance between providing the providers enough access and enough information to essentially take care of their primary machine, which is to take care of the patient while trying to lock things down and making sure a PHI does remain secure. So can you give us some examples of what really the clinical value is of the data lake that you've built there? What can you do that you couldn't do before? Yeah, absolutely. So I know the hot term that gets thrown around is real-time data. I mean, in my mind, real-time is fallacy. There's always going to be a delay between the time provider notices something and by the time they actually enter it in the patient's note. There's also a delay between communication between the clinical systems as well. But the value, which I think you asked, is really to provide them data as close to real-time as possible where they could actually see trends. Is it an issue with patient progression? Why are there delays in radiology, for example? Is there something wrong with the whole patient care workflow that needs to be addressed and providing valuable near real-time information to the key decision-makers? That's the key. That's what we're trying to accomplish and trying to fine-tune that as much as possible. Is there anything that you're asking of kind of the technology community that they're not meeting today? What do you see as the light space out there? So, if you're referring to the vendor space itself, I think the vendor market is good. We have, you know, we always tend to take the approach of is it good enough? I think it's good enough for now, but what's good enough for now is not necessarily good enough 18 months from now as the industry changes and as the organization grows. But I feel like there are good solutions and good architectures coming in play. Just like Hadoop has sort of changed the way we look at cold storage, I think there'll be new reference architectures such as that that will help us grow as the industry changes as well. Great. So what messages would you have for your peers? What do you see kind of the advances in medicine with technology in the future? I think it is, even though healthcare tends to be a liar when it comes to technology adoption, I think the whole speed of adoption is changing especially in healthcare. My advice or my counsel, if I could offer that would be don't be reactive because things are changing. It's time to look at the business and see how your team, regardless of, you know, which, what field you're in or what role you play within the organization to see how you could impact the patient care mission. RCE always says for every patient that walks in, there's 100 people within the organization that touches that patient's experience. And in my mind, I think IT and healthcare, IT folks have a big role to play in that. So that's my counsel would be to figure out how you could be that one of those 100 that could impact that patient care mission. All right, yeah. Fassi Sifisegren, really appreciate you joining with us. Thank you. Lots of interesting and innovative things going on in the medical world, of course. We'll be back with lots more coverage here from EMC World 2016. Thanks for watching theCUBE. It's always...