Peter Neupert Speaking at PMWC 2011

I run the Health Solutions team at Microsoft, and at Microsoft we're trying to build the infrastructure, the software infrastructure, to enable the learning healthcare system. What does that mean? As Brook said yesterday, it means you have to connect everything: you have to connect patients and providers, providers and providers, providers and hospitals, providers and pharma, pharma and patients, and payers, and all kinds of new care delivery models that don't exist yet. So how do you go from today's state, which is paper-based, and imagine that you start to digitize it and get data liquid so that we can use it not only at the point of care so that we can do a better job delivering high-quality, consistent care that delivers value, but also secondary use? It's a big, hard problem. What I want to do today is talk about a very practical solution that we have prototyped with the National Cancer Institute, which really, if you listen to Marty and I have all of Marty's passion around Cancer Commons and using cancer as a model, the disease cancer as a model, to activate patients, providers, and researchers to be the early point in what a learning healthcare system could look like. And what we're trying to do in the Health Solutions group is deliver the infrastructure that incrementally makes that happen without having to change everything to get started. And the learning healthcare system, fortunately I'm on the Institute of Medicine Roundtable about a learning healthcare system, where we talk about what's working, what's not working, how it might actually be applied, and what the benefits are. These are things that are relatively straightforward, but my role, as the non-doctor, non-science scientist, sort of software entrepreneur guy, is to always remind people about the patient. Because in that Institute of Medicine roundtable, the patient is absent, and they're not really thought of as somebody who can be an activator of change, a co-decision maker in a shared decision-making process that we heard talked about many times over the last couple of days, and needs to be empowered with the right sort of information, content, education, and respect in order to be a major player in this shared decision-making model going forward. And importantly, our systems aren't designed, as was discussed at the most recent panel, to get the feedback. You don't get learning without feedback, and our systems need... paper is not very good at feedback, or paper systems aren't very good at delivering feedback. Internet digital systems, patient empowerment, social network, however you want to think about it, are much better at getting feedback if the receivers are prepared to be able to take advantage of it. And so we started a long time ago, in 2006, with the idea, and in 2007 launched a patient-controlled health data repository called HealthVault. The principle of HealthVault is, at a minimum, we need to unlock the data that exists. A lot of data is digital today that doesn't get used. Pharmacy data is largely digital at the point of dispensing, at a minimum. Lab data is digital at the point of the lab result; it gets converted to paper to deliver it to your doc. Not a really good idea. But HealthVault is designed to connect to every EMR, every pharmacy system, every provider system, to make it easy for patients to get a copy of their data. Step one: get a copy of their data. And as I recall, in testifying in front of Congress and lots of other health affairs articles and other things, this was a unique approach in 2007. Today it's the law. So in a relatively short period of time, we've been able, the industry has been able, to change the expectation about the role of patients and the ability of them to get a copy of their digital data. But what do they do with it, and how does it get used by the system, is the next problem. We also have a product called Amalga, which is designed to solve a very important problem in health enterprises: most systems, transactional systems or enterprises, have lots of data that is not collected and made available for cohort analysis, secondary use, business intelligence, understanding what works. They're particularly good at doing one patient at a time. You heard Brett talk about Cash for Clunkers. Hey, it's really good for doing this kind of stuff, not really good for the future of personalized medicine. Amalga is an infrastructure that allows us to connect to all of those transaction systems and put data in a format and in a framework that allows you to connect with other providers or patients, or to create aggregates to do the kind of work that we talk about. So what we've done with the National Cancer Institute is to take those two concepts and say, how could they help accelerate research in the cancer community? As you heard Marty say, most cancer care is done in the community care facilities, not in the big 15 cancer research centers. How do we engage them? NCI has created a data format specification called PODS, here called Patient Outcomes Data Service. It's 256 data elements. We coded that up and created a way to get it easy into Amalga. A web form is one form to do it, or we can take the data out of the EMR if it exists and connect it. So the first thing is to unlock the data from those community cancer care centers. The second thing is, how is that going to work? How do we create the pull to make that happen, and an environment where once you've pulled it, enable it to be reused by the research community? And that's where HealthVault came in, which says, one of the important attributes of the cancer disease space is patients get very activated. There are a variety of different institutions, whether it's Susan B. Komen, the ones you saw today, Livestrong, who have the business or the capability to activate patients and can educate them to ask their doctor for a copy of their data. And once you have this connection, you can then start to talk about consent and all the issues about sharing and movement and making data more portable, and you can make it capable to close the feedback loop. Because one of the important things about cancer treatment is, how are you doing? And how do we capture that data and make it easy for the patient to be engaged in the co-production of information that's not only delivered at their point of care but enriches the research infrastructure, and make it readily accessible for those that are doing research. And we've built a couple of screens; it's a whole application. And the way both these models work is both HealthVault and Amalga are fundamentally application platforms, which is what Microsoft builds: application platforms to allow for the creation of a variety of applications that leverage a data model and some data capabilities and make it reusable. And these are mockups about how one might interact, how a patient might interact relative to their treatment. And then this is a 28-page slide deck that I condensed into three to demonstrate that this is more than a concept, to demonstrate that today there is the infrastructure where we can start to do this, and there is the opportunity to work together to start thinking about how can we create a feedback-driven system with the patient in the center that enables not only our community cancer centers to participate in the framework, but enables that to go up to research, primarily academic research-driven NCI, caBIG, all of that other infrastructure that already exists. What we've tried to do is plug into those infrastructures that exist, make it easy at the margin to start moving data, to encourage and empower the patient. And what Marty's done a great job on, and we hope to find a way to collaborate, is to add the extra piece: how do we get the expert content and make that accessible both to the provider and the patient, and really complete the loop. And we hope to be able to figure out ways to do that. We have demonstrated in a couple of different environments where engaging the patient in a chronic disease case, two different tests that we did, pilots that we did, one with Cleveland Clinic, one with Kaiser. Kaiser was hypertension; Cleveland Clinic was both hypertension, CHF, and diabetes. And it was a simple hypothesis: if we engaged, getting biometric information, this is off a device direct into HealthVault, direct into their EMR, could we change the outcomes simply? Could we improve the intermediate outcome that they were trying to drive with fewer office visits? And the answer was yes. It's not that complicated. We can do this today. We talked about the notion of the reimbursement systems needing to be adjusted to encourage this kind of behavior, because one of the bad outcomes for physicians is fewer office visits is lower income. It's not a very... it's a really good idea, it's better care, but it's lower income. So we have to fix that. And there are these Accountable Care Organization concepts which will hopefully address that. But this is stuff that can get done today with off-the-shelf hardware and software. And we hope to encourage the concept of engaging patients in their own care and making that part of the standard of care, standard of practice. With that, you can start to imagine if you build your information systems, as Brett described earlier, so that you have not only good systems for point of care but also systems that allow for secondary use, the creation of registries, the creation of cohorts, the creation of understanding, we can create the learning health system which can transform not only what we're doing with point of care but clinical research. And if I heard the bell, I will stop with that. My goal is to recognize and champion that patients have a pivotal role to play and that they're prepared to play it. I think many in the industry think they're not, and I think they are. Thank you.