 This will be about 11 minutes. So thanks. I'm Ken Kawamoto. I'm from the University of Utah. I was recently recruited there to direct the operational clinical distance board knowledge management, obviously not just for personalized medicine, but for all traditional medicine at the health system. And I'm also a faculty member in biomedical informatics. So before I begin, I just need to give a few disclosures. I am or have recently been a consultant to partners in Flexion and RAND. I also was formerly a consultant for a relagent and formerly a co-owner and consultant for a company called Clinica that's in this space. I have no financial competing interests related to open CDS, which is the open source technology I'll be discussing today. So today my objectives are to first describe what clinical distance board or CDS is. I think this is an audience that probably is familiar with the term, but may need a little bit more information on it. And I want to specifically talk about how it can help bridge the translation gap for personalized medicine. And then I'm going to identify requirements for scaling distance support for personalized medicine on a national scale, and then describe an open source standards-based distance support resource known as open CDS that through actually funding from the NHGRI through a current development award I was able to start for enabling distance support for personalized medicine. So I'm going to begin with a definition. So clinical distance support can be defined as the provision of pertinent knowledge and or person-specific information to clinical decision-makers such as patients and clinicians to enhance health and healthcare. And by supporting evidence-based decision-making, clinical distance support can support the translation of evidence into practice. So for example, when we looked at clinical distance support systems that have been evaluated through randomized control trials, we found that when they provide patient-specific care recommendations that are automatically provided within clinical workflows, they significantly improve care quality in over 90% of randomized control trials. So just to give you a sense of what people typically mean by distance support in this field, here's an example, disease management dashboard that I created while I was back at Duke Health System. So this is a disease management dashboard available to clinicians as a part of the electronic health record system. So when you pull this up, for example, here for diabetes, it'll show you care metrics that the American Diabetes Association says you need to worry about. It pulls in the relevant patient data when things were last done, the last values were, and a bottom line, very actionable recommendation on what you should do. So this was very well-liked, used very widely at Duke. So when you think of what this kind of technology can do for personalized medicine, I think there's a few points where it clearly makes sense. So for example, for test sorting, you can definitely recommend an indicated test. So as clinicians are working, you can imagine that you would get some sort of reminder or pop-up or whatnot that says, hey, did you know that for this type of patient, you should be doing this genetic test to make sure that you don't get an adverse event? Or perhaps more importantly, you can recommend against a test with dubious clinical utility. So you can imagine a clinician is about to order a test and you know that this shouldn't be done. You can say, well, you know what? You probably shouldn't be doing this and this is why. You could also personalize test interpretation. So instead of getting a genetic test result back that says, here's the genotype, and in general, this is what it means. You can imagine that it can be combined with clinical data to provide a personalized risk assessment and care guidance. So the key part there is you need to pull in the clinical data at the same time to provide that guidance. Also as our next speaker, I think we'll be talking about, you can clearly make use of family health histories to, for example, tailor preventive care guidance for cancer prevention. And of course, you can manage therapy. So for example, distance support can help you provide genetically informed pharmacotherapy guidance. So this is an example of a Warfarin Pharmacogenetic Dosing System that I developed while I was at Duke. So this is for a commercial order entry system called Nikesen. So you can see here, you collect a number of relevant clinical data. Also identify the patient's relevant genotype. And then other relevant factors like previous doses. And based on that, you provide very specific care recommendations on what the dosing should be for this patient for Warfarin. So I think some of these examples show, I mean, I think intuitively and based on review of the evidence, this approach is very effective. It can really help provide better care. The problem with this approach is that distance support has been very limited in its scale. So despite decades of demonstrated effectiveness of these kind of tools, distance support is not widely available. So for the practicing clinicians here, unless you happen to work at one of the leading institutions for developing these kind of systems, you typically have fairly low or no distance support available in your clinical practice. So this happens because most distance support is designed to only work within specific institutions in health IT systems. So it's great that you have a paper in the New England Journal showing that this kind of approach for distance support works, except 20 years later, most people still aren't able to do that in their clinical practice. So we still are trying to scale up traditional distance support capabilities that have been validated for decades. My main point here is that unless we focus on scalability, distance support will have a limited impact on personalized medicine for decades to come. So here's an example. Here's a New England Journal paper called Protocol-Based Computer Reminders, The Quality of Care and the Non-Perfectibility of Man. Great paper, randomized control trial had several hundred rules on disease management, adverse event prevention that showed this improves care. The notable part about this is it was published in 1976. So this was published before I was born. Right? So this is a, and we are still trying to replicate this capability across most health care systems. You look at the rules here, most health care systems do not have the rules that were validated in this randomized control trial in the 1970s. So the question becomes, what's needed to scale distance support for personalized medicine? So we published a paper on this topic that where we thought through this, and I'm particularly interested in the space also because I co-chaired the distance support work group for a group called HL7, which is the standard setting body for health IT. So there's at least three main components. So one, you need to standardize the representation of relevant patient data. If people talk about diabetes differently, medications differently, or observations differently, you're never gonna scale this. So this is for both clinical and genetic data needs to be standardized. You also need centrally managed repositories of high-quality computer-processable medical knowledge because if you don't know what you should do, you're not gonna be able to do anything. And you also need standard approaches for leveraging this knowledge to provide patient-specific advice. So what I'm gonna talk about briefly now is a tool called OpenCDS, which is an open source, standard-based, freely available distance support platform designed to enable distance support at scale. So through funding from the NHRI, I was able to start this initiative. It's certainly intended to support and address the requirements for a national distance support infrastructure for personalized medicine, but it also encompasses more traditional medicine as well. So some key components of this, it uses standard data models. So I won't get into the details, but a lot of my activities over the past several years has been defining and getting adoption of some of these models as international standards. It can also leverage various knowledge resources. So looking at all these different knowledge resources that are available, it can leverage them both when they're externally internally developed. And it provides a standard approach for EHR systems to leverage these kind of resources over the internet through what we call web services call to obtain patient-specific advice. So this has grown into a fairly large collaboration. There's a number of collaborators from both the public and private sector and actually a fair amount from outside the US. And these are all folks who are interested in pursuing a standard open source approach to creating the infrastructure for this kind of distance support that scales. So without getting into technical details, I'm just gonna show you how this kind of approach works. So you have a very simple diagram, given institution with an electronic health record system and their data sources, and then you have open CDS as a care guide and service. So what you can see here, for example, is this is the Duke electronic health record system in this kind of approach when a clinician clicks on, oh, I want the disease summary for this patient. Patient data is collected, sent to the service over the web, of course using standards. And then again, using standards, you get care advice that's generated, return back to the system so you can do things like this disease management recommendation screen I showed a little bit earlier. Here's another example. So as I mentioned, this approach isn't restricted to particular types of knowledge representation. So here's warfrendosing.org, which was mentioned yesterday, and I think a lot of people are familiar with. So it's a website that provides genetically guided warfrendosing. So if you'll notice with the screenshot I showed earlier, it'll say here, it says dosing guidance received from warfrendosing.org. So we showed how we can use the standard wrapper and they can provide this information. And just to show you what we have in terms of the actual web-based authoring platform, so without going into details, I'll just show that there's a web-based authoring for example, human readable decision rules, decision tables, so for example, this is for a vaccination schedule and this all allows you to provide these kind of recommendations. And our latest addition of, for example, decision diagrams. So you can see here, this is how we're encoding family history, risk analysis for identifying if a patient that increases for BRCA mutations. So just as a status, we are releasing a 1.0 release later this month. We already have a beta release that's available to interested parties and we have a multiple ongoing initiatives. So I'll close with some recommendations. So I think it's really important to make this in support, a core component of the personalized medicine vision. I think without it it's gonna be difficult for clinicians to make sense of this. We also need to consider and prioritize scalability for distance support. So like I mentioned, if you don't consider this, we'll have nice funded, published New England Journal papers in 2015 that shows this is a great way to improve patient care in 2050, which is about how long it's been since that New England Journal paper. We'll still be trying to figure out how we do it in most healthcare settings. You should leverage also relevant resources, open CDS is one I described and there's multiple other efforts I didn't have time to get into that are specifically focusing on this area, just not in the personalized medicine space that could be leveraged. Also as a note, I think it's important to align with an influence on EHR Meaningful Use Regulations. So for those who aren't familiar, this is the regulation that is driving through various reimbursement models, EHR, FOSI in this country. So in order to have an impact in the space, it's really important for I think the NHRI and relevant parties to really influence that. And I think as a final point, we need to start building this infrastructure now. If we don't do it, divergent and compatible approaches will develop and it's actually a really good thing that a lot of this has not yet been developed because one of the more asses in other aspects of traditional clinical medicine in the space is people have done so many divergent things that it's really hard to get everyone on the same page at this point. So I'd like to acknowledge the NHRI for funding me as one of those few dots at the end of the spectrum. The University of Utah Department of Biomedical Informatics and the Office of the National Coordinator and also a number of collaborators. If you like information, opencds.org has all the information. Basically, if you have a Google-based email access, I can give anybody free access. And I'd like to thank you for your attention and take any questions. I had a question. Are you interacting with some of the large EHR providers now like Epic and some of the others that seem to be? At the University of Utah, we're working with Epic and Cerner installs. There's also a collaborator from a health system called Mainline Health in Pennsylvania. The CMIO, the Chief Medical Informatics Officer there was the head of Disinsported Siemens and just got a contract with them to try to integrate this kind of technology. We're definitely interested and we will certainly be doing it for Epic and Cerner because that's my operational duty to do that at the University of Utah. Okay. And can I just ask you, was it easy to get Epic and Cerner interested in what you're doing and involved or not? Let me preface that to say, I started in August at the University of Utah so we are still getting started. But just as an anecdote, we had some collaborators from Kaiser of Southern California who wanted to work with us because they were having difficulty having Epic do things for them. And so I'm under no illusion that a company that's selling billions of dollars of EHR systems without making any modifications will be that interested in changing the approach they do things. So the jury's still out but I think that's why engagement in the meaningful use process will be very important because as part of some other work I've done I've interviewed EHR vendor executives and they basically say if it's not meaningful use we're not interested right now. Can you show your slide again where it has like with the cancers in the breast cancer? This one? Yes. I need to make a preface there that this is still under development because it uses an open source technology which became available a month ago. So does it ask anything about if there is a male relative with breast cancer has it been tested and if so what the mutation is to be able to take it? I just took the straight US preventive services task force clinical guidance. So none of this, all this is only as good as the source of the recommendation. Yes, sir. Is there a library on the open source server site of these decisions? Is this, does this become something that the community can access or we can with and understand? Certainly the infrastructure is all open source because we've been focused on infrastructure and as I noted we're doing the 1.0 infrastructure release this month and actually some of the base standards on this we just finalized as a standard in September. So we've been focused on infrastructure but we will be making a lot of the content open source. For example, we have a collaboration with the New York Citywide Immunization Registering the Alabama Department of Public Health where they will be providing all of their immunization related distance support all available open source through open CDS. Hi, I'm Mark Hoffman with Serner and great to hear about your work. I think there's two perspectives that at least within our company we have. We feel that clinical decision support is real within commercial EHR systems but at the same time we recognize the challenges of exchanging decision support logic across systems and also recognize that the maintenance in the current model of decision support logic is very cumbersome and so we are working on ways to get past that burdensome maintenance of rules and at least speaking for Serner we see initiatives like yours that provide open source thought capabilities as part of how we can work with the community to move forward so I'd be happy to talk with you afterwards. That'd be great and I'll just mention Serner has a wonderful capability called M-Pages which allows you very easily to integrate this kind of content so right now my headache is primarily on the epic side. Yes. Oh, I think I'm out of time. Thank you.