 Live from Cambridge, Massachusetts, it's The Cube at the MIT Chief Data Officer and Information Quality Symposium with hosts Dave Vellante and Jeff Kelly. Welcome back everybody. We're here in Cambridge, Massachusetts. This is Dave Vellante with Jeff Kelly. This is Silicon Angles The Cube. This is our flagship program. We go out to the events. We extract the signal from the noise. James Noguz here. He's the CIO of Partners Healthcare based here in Massachusetts. Jim, thanks for coming to The Cube. Appreciate that. So first of all, welcome. Second of all, you have a, well, let's talk about your role first. Partners is a very well-known organization here, but our audience outside of Massachusetts may not be as familiar. Yeah, so partners, its founding members are Mass General Hospital and Brigham Women's Hospital, which, combined in terms of NIH funding, have the most NIH funding in the country. It's over about a $1.5 billion in research funding. So in this particular conference and things like big data, big data is extremely important to healthcare as we talk about improving patient safety, really cutting costs, but important to our academic medical centers is discovery, just massive amounts of data with genetics and genomics, with the digitizing of images, is a lot of very new opportunities for discovery, both asking questions and sort of finding answers that we weren't looking for and things like pharmacovigilance. So it's exciting. Well, this is a, Jeff and I, this is a topic that's near and dear to our hearts. As you're saying off camera, his wife's a nurse. I've followed big data for a while now, and Jeff Hammerbocker of Facebook and Cloudera are very famous for saying, the best minds of my generation are trying to figure out how to get people to click on ads. That's what big data is becoming. He left Cloudera. I guess he's still evolved, but he's now at Mount Sinai Medical and doing some great work down there. So that statement was sort of tongue-in-cheek, and it's becoming not true. Right? The best minds are actually solving some of the world's most difficult problems in a variety industry. So I wonder if you could talk about big data in healthcare. You've got a session tomorrow on that topic. Where are we at with big data in healthcare? A couple of places. One is, we have massive amounts of what I would call administrative claims data that we can start to model in terms of what our costs really are and really taking on the task of spending total medical expenses and lowering them, because it's really unsustainable, the current growth rate of medical expenses. So that's blocking and tackling, but just being able to go through massive amounts of data and do that analysis. More importantly, I think, is looking for patterns in clinical data in terms of outcomes that drive evidence-based medicine and clinical decision support, because what we have found over the years is people talk about clinical decision support, and they may embed thousands of rules. And some of these rules can be irritating to physicians like, why am I getting this alert? The fact is, I think now that we have big data, we can go back and look at the efficacy of those rules, because we actually think probably 60-65% are unnecessary and don't really result in a different treatment protocol or better outcome. To get in the way. Yeah, exactly. So that evidence-based medicine. And things such as pharmacovigilance, I don't know if you remember the BIOX, what happened with BIOX? I was on BIOX at the time. Okay. Yeah, from my back. I do remember. We're fairly certain, with big data analysis, that that pattern could have been picked up and predicted well in advance of it being reported and sort of their being an intervention. Because it ends up in clinical trials, while they're extensive, they're still fairly narrow in terms of their patient population. It's really once a drug is out in the field is being able to not just, let's say, partners that has three and a half million active patients, but being able to look at outcomes across multiple medical centers and being able to quickly assess, could something be happening. So the whole idea of pharmacovigilance, and then in terms of treatment in genetics and genomics, what we're finding is with different variants, and this is especially related to the treatment of cancer is particular cocktails may work extremely well for you, but through genetics we may find out that it's actually harmful for you. So it's really being able to do that mass customization of treatment protocols, whether it be for cancer or other disease states. And that's what really big data is starting to offer us. And then being able to look at digitized images and do an analysis on a massive basis. People on the panel tomorrow, Dr. Sean Murphy, he talks about the fact that now with digitized images they actually think, you know, diagnosing a schizophrenia in adulthood, that actually going back to childhood and being able to go through images quickly, because kids get concussions, they have MRIs, that they can look for other patterns in childhood that can be predictive, schizophrenia in adulthood through image analysis. Some amazing examples there. I mean, the first one you're talking about, the Biohux example, I mean the trials are samples, and sampling just doesn't give you the efficacy of the data in the field. And you talked about, you know, basically the cocktail, it used to be brute force hit and hope, yes, to see what worked. My question is, there's a big theme in the big data world about real time. Is real time seeping into your world yet? Not yet. Do you see it happening? I think it will happen, but it isn't quite there. More isn't that retrospective evidence-based analysis, which is why you're seeing big data. I think first take a foothold in the research community in that area of discovery. But I think it can. We've talked about also how to social media play into the health record in terms of big data. Can you look for patterns and help postings in that that are indicative of possibly a person slipping into depression? It ends up that depression has this effect that when people become depressed, they stop taking their meds, not their depression meds, but all that it really impacts healthcare and being able to do that quickly is extremely important. So technically, how are you attacking this problem? Is it massive Hadoop clusters? Is it a combination of EDWs and additional technologies? Well, one thing we know, it's not a monolithic database. Not a god box. No, a lot of distributed data, a lot of metadata, and we have the traditional EDW that we're actually even doing that with what I would call a federated model with metadata and creating data marts. But on the research end, in fact, we are bringing in like Hadoop clusters and starting to really work with those and say, how can we leverage those in terms of big data in healthcare? And big data, not to slight it, but it's just more data. Just massive amounts of data is really big data and some of it is structured. But in healthcare, the majority of data is actually unstructured, so the dictated notes of a physician. Now we use things like NLP and that to try to do analysis of those notes and put it into some type of structure. And same thing on images. Although images now, they're starting to develop what I would call header information along with the images that's giving it some structure. Yeah, and you're right, big data, it's more data. But it's also new techniques of handling it. And as we talked about before, it's not taking little samples. We're seeing the outcomes all throughout healthcare. We swipe a credit card. Somebody tries to rip us off or steal our credit card. You get a phone call virtually immediately. Those things weren't possible five or seven years ago. That's correct. And the other thing where it becomes really important, I talked about genetics, so just the massive amount of data. As there's new knowledge, being able to quickly go back and apply that new knowledge to data that you already have. And we think that's actually going to probably raise some ethical questions and some privacy questions in terms of what do I do when you had a genetic test for me five years ago. Today I discover something and I have to call you up and say, you have a really bad disease. And what is our responsibility to do that? To go back through that massive amounts of data and do that type of analysis as there's new discoveries and new knowledge. But in the epiphany there is this real value in that historical. They used to be, oh, the data's old. It's not valuable. We'll keep it just because, but sticking it in an iron mountain somewhere, hopefully we'll never have to go get it. Now it's a valuable source of information. Yes, and petabytes, you know, every month. It's big that we're generating it. So you've got ethical considerations. But talk a little bit about the challenges of working with large-scale data analytics in an environment, healthcare that's highly regulated and somewhat uncertain. We just saw yesterday two different courts come to different conclusions about one aspect of the Affordable Care Act, whether it was valid or not. Correct. So you've got, the sand is shifting beneath your feet pretty much all the time. How does that make your job in terms of managing and making better use of data more difficult? Well, I think what it really points out is having agility, which as you said, no God boxes. Right. And the ability to apply, whether it's Hadoop or other BI techniques to data. And the fact that, yes, in fact, and I don't think it's isolated to healthcare, but knowledge changes over time. And applying that new knowledge to that data is what we need to be able to do fairly quickly. I don't know that I've fully answered your question. Well, no, I think so. I mean, it strikes me that when you've got a regulatory environment that's changing so rapidly that as a healthcare organization, as any organization in an environment like that, you've got to be flexible for sure. Yeah. So let me follow up. Sure. So you're right on the regulatory means in terms of there's fairly extensive consenting process. You have to obtain consent from patients to use their data in research. And so it's really, culturally, it is being able to communicate to patients the value of that research while maintaining their privacy. So in healthcare... Get it to opt in, essentially. Yes, you do have to opt in. And we're actually doing that now. And giving it incentive to do that, right? Exactly. And the incentive is improving healthcare, still respecting the privacy of the individual. So if you've ever heard the word HIPAA, there's a lot of regulations in terms of security and privacy. In the research community, there's a whole institutional review board that you can look at cohorts of patients anonymously, but once you're actually going to start the research, you have to work through the physicians, contact the patients and obtain that consent. It's a little bit laborious, but I do think it's necessary in terms of respecting the privacy of individuals. And you've not only have to get the patients to opt in in a situation like that, but in terms of getting your clinicians, your doctors, nurses to actually start using, or start using more data-driven approaches to making decisions, whether it's developing treatments or diagnosing patients or day-to-day care of patients. Talk about that. How do you actually go about that change management component? And doctors are notoriously a little bit fickle about how they operate and sometimes think that they know better than the data might be telling them. Perhaps maybe that's changing, maybe that's not a fair assessment, but how do you go about actually getting, in your case, clinicians to actually adopt some of these new approaches? Yes. So we've had good adoption with our electronic health record and clinical decision support, but we've been doing it for some time. And I think it's when they start to see the value when you can produce statistics of how many drug-drug interactions didn't occur because there was clinical decision support. And we can track that and say, this order was going to be placed and we prevented an adverse event. Well, that becomes meaningful to clinicians as well as the clinician is no longer sort of a single person. It's really team care. And I think they realize that some of that has to happen asynchronously and that's where technology and decision support can aid sort of that asynchronous treatment of a patient. It's just not, you know, the single physician in a sense they're the quarterback, but they have a whole team behind them of nurse practitioners, physician assistants, the allied health professions, pharmacists, all those come into play. So Jim, I wanted to ask you, the big theme of this conference is the chief data officer, the emergence of that role, certainly big in regulated industries like financial services and health care and government. What are you seeing within your industry as far as the CDO? What does that mean for the role of the CIO and how is that emerging? So I think in health care what I'm starting to see is an affinity for people saying, we need chief data scientists and they're different than, you know, your typical, what I would call, business analyst or people that do research against the EDW. So there's recognition of that. In fact, we've talked about the concept of medical research. We have things that are called core labs that people can reference, that you have a group of chief data scientists that really are the experts and understand the data that can help people sort of when they want to do their big data analysis. It's that old figures, li-liers figure. All right, let's get the question right and let's get the right data set to get the appropriate answer so that I don't have conflicting conclusions coming out of the same data set. Now, those chief data scientists, do they report to a chief data officer? Don't know yet. I would say we're not there yet in terms of a chief data officer. We are there in terms of enterprise governance and sort of that concept of having a core lab. I think it will evolve into a chief data officer and you asked about the CIO and I say this and maybe because I'm near retirement I think in ten years the role of the CIO may be dramatically different. In fact, there may not be CIOs because I really see what I do on a daily basis is the CIO has to maintain the infrastructure. A lot of things are moving to the cloud. A lot of the things with converged technology there's less in terms of sort of managing that complexity and it's becoming a utility. So, in the 1800s, every manufacturing plant had its own power plant and yes, we're still building data centers but I think you're going to see less of that and I think the CIO, those that are going to be successful need to focus on sort of what I would call the optimizations of workflow processes whether that's LEAN, Six Sigma, I mean there's many programs but I think the focus is going to be optimizing the business and at the CIO, I think typically you heard CIO say we're an enabler of the strategy and I think the CIO needs to play a role in terms of the formation of the strategy in a sense they're almost morphing into chief operating officers so I could see the day when a COO is fairly competent in information technology as well as business operations and has like a chief data officer, a CTO reporting to them but there really isn't the concept of the CIO anymore it's more a convergence of the COO and the CIO role. Just the theory. Yeah, so it sounds like more focused on business process optimization versus IT maintenance, keeping things up and running. Correct. Yeah. That's interesting. You know, when I was a kid, you know, I got out the timing light on the car and checked the points. You know, do it anyway, right? Exactly, you drive the car. I don't go under the hood anymore. Unless you really know what you're doing. It's probably good analogy, the CIO is coming out from under the hood. Yeah. Wow, that's well, it's interesting coming from a CIO that you have that view. So, the follow up question would be and we've been asking a lot of folks is the CDO role a flash in the pan? You're saying no. We've been asking people, will the CDO role be here in 10 years and you're saying absolutely. I think it will be. I think because not just big data but analytics, I think really is a competitive advantage in a company that doesn't invest in it may not be here in 10 years. That's really what it's, I think analytics is really the next frontier in competitiveness. And you know, you made some references, Google everybody in terms of getting people to click on advertising. You know, that's what it's all about. Well, you remember. Not in healthcare. But you remember the early 2000s in the car, does IT matter? Clearly IT matters. IT matters. But IT is so now embedded in the operations. And you know what I find, at least in healthcare, I run into physicians in particular aspects of IT. Hey, they know it as well as I do. So, there isn't a knowledge advantage anymore I think in the role of the CIO which is why it needs to evolve into what I would call the business operations and that optimization aspect. Well, we were at the service now of a knowledge conference earlier this year in May. And Frank Slutman stood up and he said the CIO role, essentially he said it differently than what you just said. He didn't say the CIO role is going away because all his customers would have stabbed him. But he essentially said that the role of the CIO must morph into one of a business person. Which is exactly what you're saying. It's a COO role. It's a business role. And you're going to have a CTO that worries about all the architecture and keeping the lights on. Making sure that it's all working properly. And that's not going to be his or her role to be a data czar, right? That's a different emerging incremental role. And I think CIOs rather than see this as threatening should see it as exciting. Yeah, because they can get a promotion of COO. That's a new way of doing it. When you're sitting with senior leadership you're no longer at the end of the line. As decisions are made you become part of that formation of the strategy rather than the receiver of the strategy in that oh no moment of how am I going to do this. We're talking growth path for CIOs as well. A lot of CIOs aren't deep techies especially in large organizations. But in smaller companies the CIO oftentimes comes out of the technical role. So there's your path. You go be CTO or a business operations role, correct? Absolutely. So if you're an existing CIO how do you start who maybe doesn't have a CTO at the table right now how do you start to kind of wedge your way into that role and take more of a business side role versus just the infrastructure role? I think you have to actively engage senior leadership. You may even have to push a few doors open rather than waiting for them to be opened. My senior leadership has been very receptive of being part of the leadership team and I always really push that I need a business owner for every initiative that IT should not love an initiative any more than a business owner usually it fails then. Is there no such thing as an IT project in your world? Well there is an infrastructure so I do feel that I have the responsibility for infrastructure I need to be able to look around corners and say how might this technology apply to our business but in terms of application support of the business processes I really want it to come from a VP on the operations side of the house or a chief medical officer or a clinician and not IT but being part of the senior leadership team then you can introduce ideas as well as any other senior leader and I think the tough part for CIOs and I would say even sometimes I have a level of discomfort starting to comment on non-IT issues from a business strategy perspective but you have to do that to be part of the team. Sorry who do you report to? I report to it's an executive VP that reports to the CEO and he basically has all the CX's reporting to him. Okay so there's a layer beneath the CEO that is the buffer. The heat shield to the CEO. I mean on a weekly basis I'm part of what we call the operations head and it's the CEO and all the leadership. Yeah sure the CEO's involved. Yeah exactly. Okay and you mentioned the power plant and I have to ask as we're talking about Nick Carr's IT matter he wrote another book called The Big Switch what about the cloud in your world? Will privacy and security concerns keep it out of your world or is that just an illusion? I think it's a bit of an illusion in terms of I think people are concerned about privacy and security. I mentioned HIPAA there is something called business associate agreements that a cloud provider takes on the responsibility and for many years some of the cloud providers said no we won't sign a business associate agreement we don't want that liability. We signed a business associate agreement with Amazon. I was going to ask about that. So we're starting to move some of our research computing up into the cloud. I mean we see Amazon being very aggressive in terms of making itself compliant in f-care and government the CIA cloud etc. We see no reason why at their growth rate at their innovation rate they can't provide that level of security and privacy. Do you agree with that? I agree with it and they can do it better than I can. It's similar to security is something else I worry about. Having to manage system security provider makes sense to me because I can't look at blogs that are generated every two weeks. I think it's fair for organizations to say Amazon security is one size fits all and I can't fit into that size. That makes sense to me but to say they're not secure which a lot of the competitors will say it's dangerous. No that's not the case. It's maybe not as flexible yet. Over time Amazon is going to keep building in others but really Amazon today building capabilities that are going to satisfy a bigger and bigger space. You live in this world you hear it from your colleagues oh cloud is a bad word it's never going to happen but you're admitting that a company like Amazon or Google ultimately is going to have better security than you could ever architect because they've got the engineering resources and the PhDs running around that's what they do. What percent of the organizations do you feel are of that ilk? Do you think it's of the vast majority because a lot of companies will say no way I can do it better particularly for instance in financial services do you think again is that an illusion because you're willing to break eggs here so this is good. I'd say it's probably 30 to 35% that think like I do and I think the other 65% are still proceeding very cautiously but I don't know if it's more about security or it's more about turf. Hanging on. Yeah exactly. There are some risks in the cloud I mean in terms of obviously connectivity and what happens if there's a disruption to connectivity or distributed denial of service attack that shuts down your bandwidth you know out to the internet so there's risk but there's risk everywhere. Yeah well there's also the impact whether it's symbolic politically if something bad happens and you've gone to the cloud then the backlash is C whereas it happens to internally whether it's Fidelity or TJ Max or you know but increasingly that's not becoming the case I mean it happens internally you're still getting a lot of the CEO of Target is gone so it's a factor I think but I don't think it's as big a factor as it maybe was a couple of years ago. Absolutely. Now I just think you're seeing more things that are subscription based I mean if we're all going to I don't know that will all be a lot of people will be on Office 365 right? Yeah or Google Docs or Google Docs so I mean it's moving that direction and it's the business model you're even going to get seen with you know in the SAPs the Oracle Oracle talks about its cloud SAP we had them on earlier we're all about the cloud Yeah exactly because they realize you can really take a lot of cost out of your organization moving things to the cloud and not having to support that infrastructure Okay so as a CIO who's essentially come on the cube and said look the days of the CIO as we know it may be numbered I'm really soft peddling because I don't want to put words in your mouth but that's essentially what you're saying is that the role is changing and it's going to transform and it may in fact disappear in a lot of cases and organizations and be streamlined to COO CTO and CDO so what advice would you give to your CIO colleagues who aren't necessarily going to be retired in 10 years where should they be focused? I think you embrace it and start planning for that transformation because I really think that will be the new role of the CIO Maybe it'll retain the title but regardless I think it'll be as I've described in more of a COO role Functionally yeah Alright Jim we'll leave it there Thanks very much for coming to the cube Alright keep it right there Jeff Kelly and I and Paul Gillum will be back right after this word This is the cube we're live from MIT and Cambridge Mass We'll be right back