 So I want to give you acknowledgments. What is it? It's a EMR that provides access, and the keyword is access. All we want to do is to be allowed in. Once we're in, we don't ask the EMR vendor to do anything else. We provide the capabilities that we've heard about today and that are illustrated on this slide. Now, we've heard and we've seen the previous presentations. You can do a lot of these things, the collection transfer integration, the delivery of support. But some of the things like the patient empowerment, the better understanding, the family social networking, these are things which are really not done well within EMRs. And we think it's better if we think about bringing applications from the internet into an EMR to do that. So we have developed a relationship. We got permission from Epic. In March next year, Epic's 2010 release will enable health heritage. Anyone who buys Epic's 2010 release will automatically be able to, if they desire, to turn on this application, which is external to Epic, but which has a door into Epic. What are the principles of this system? It attains a limited amount of essential data from the primary sources. We started with family history and cancer risk assessment because you'll see we were funded by the NCI, but we plan to go obviously beyond that. We understand what the data represents. We input the data in a way to either new or existing support tools. We deliver it in ways that people can understand and then we constantly monitor, approve it. We've heard this throughout this meeting today. This is what clinicians want. This is what consumers want. As we said, we are supported by the National Cancer Institute in this initial release, so we concentrated on all the common cancers and the syndromes that are associated with that, linking family history to these cancers. The nice thing about that is family history also serves as the foundation, as you all know, for genomic medicine beyond cancer and in all the other specialties. Let's look very briefly at each of these principles. We obtain a limited amount of essential data from the various primary sources, including EMRs. With our Epic agreement, what we're able to do is we're able to go into the Epic EMR, pull out medical, surgical, family history, laboratory data, and we're able to pull that into our application and we're able to return to Epic risk results and recommendations and pedigrees to the physicians and to the patients who are using MyChart, which is the tethered PHR, which is associated with Epic. There's a single sign-on. A patient activates this service. This is a patient-oriented decision support application because we think you need to involve patients from the very beginning. They sign-on to their MyChart aspect of Epic. They're automatically registered then into health heritage. They have to go to health heritage and say that they want to use it. They request from health heritage the ability to pull their medical history into their chart from Epic, their documented medical history, when they had cancer, if they had cancer, the age, and the type that it was. If they had a colonoscopy, how many polyps did they have during that colonoscopy? All that information is pulled into their medical record and then we encourage them to say, okay, let's go to their family members and have them do the same thing and in the process we will create an electronic family tree. So we have a family tree which is populated with data from the primary sources in the record, authentic data, and which is integrated in a comprehensive way. That's always been a barrier. We have system interconnectivity. We, providers access, can access health heritage from hyperspace within the menu or toolbar within their Epic system and they can see both provider and patient context. So there's transparency across the different systems. Data exchange, what are we doing? We are moving from Epic data which is in their problem list, their medical history, their patient demographics. We're moving data which is in textual format path reports, operative reports, and we're taking those results and we're moving them into health heritage to populate an individual's personal medical history. We are then returning to Epic, which is an HL7 message, a narrative text containing what the recommendations are and we're depositing that as a test result into Epic that the clinician can look at directly. How do we know what we have when we pull this data from Epic? And this is a very important, I think, concept. We initially took the information that we needed to do family history and risk assessment in those cancer syndromes and we mapped them to UMLS, SNOMED concepts. Then we took what is a constant roadblock in this field which is Epic has a problem list or a diagnostic list which is maintained by a separate vendor and it's a proprietary system that has no relationship to anything in the world. So you automatically say, what am I going to do with this huge diagnostic master file that is linked to nothing? Well, what you do is you map it to SNOMED and that capability is increasingly there and then what we did is we used the NCBO bioportal to look at the parents, those SNOMED codes who are in the Epic diagnostic file that were also needed to be in health heritage. We identified 4,700 so that now when we make a web service call into Epic and ask for this information and if one of those SNOMED codes that we're looking for is present then we'll set the attribute that we need in the decision support system in health heritage to present. So we've overcome this barrier and increasingly being creative you can go into systems that have proprietary data formats that are not standardized and you're able to achieve understanding of what that data is. Now, we input that data into existing or newly designed decision support tools. We heard that decision support tools are not used. There are thousands of them out there in paper that are not used because they're not connected into the system. In many cases it's not necessary to design new ones. We just want people to use the existing ones with the appropriate data. For our purposes we have a very good existing one. The NCCN guidelines, there are hundreds of pages very detailed, wonderfully precise guidelines. They're completely unusable by any human being because they're so long and they're so detailed. So what we did is we made them computer friendly. We took and we developed decision trees based out of the current literature. We just didn't use NCCN. We updated it where we thought was appropriate, although it keeps itself updated really, really very, very well. And we did a decision trees for all of the specific cancers and all the specific syndromes. This is a substantial undertaking. Fortunately it only needs to be done once, but this is just the top of a decision tree here that actually if you would print it out it would go about 16 feet long in order to take all the different permutations and possibilities into it. But what it does is it automates it. You don't have to go to those hundreds of pages of guidelines. It automatically takes the data that we've collected. It puts it through the NCCN process and it comes out with those specific risk and recommendations. And it delivers the output in ways that can be understood, stored and easily shared by consumers and clinicians in trusted ways. This is something else that we've heard from a lot of people that they want to be able to do. So after we process the system we collect the information from individuals and from their family members with sufficient precision to be able to do a risk assessment and recommendation. We can put people into risk stratification. And we're trying to look at both ends of the spectrum. We're trying to recognize that there are many millions of individuals who because of their individual patient characteristics really don't need routine screening with the frequency with which it's currently being done in this country. And we're trying to suggest to some of those individuals there may be the ability to reduce the frequency of their screening safely. At the other end we want to make very sure that we identify very early and very soon those individuals who indeed because of biological or hereditary predispositions are an extremely high risk for developing problems. And they are identified early in the process frequently through family-based structures and analysis of the family history and that they're given the appropriate advice early on to be able to access the services that they need. We present that report about what they should do to the patient and then we ask the patient whether they want that report to go to their clinician and which clinician they want it to go to. It's a very important process of patient control. It's not how things are traditionally done. Traditionally things done, test results show up in EMRs and the doctor decides when the patient can see it or it defaults to having the patient see it after a few days. We think this is patient information. It's their family information. After they know what the NCN recommends for them then they decide who they want to send it to and that test result then goes to the clinician. They have a number of links that they can go to when they receive that result. We also are building in crowdsourcing which is trying to link them to sites like patients like me where they can exchange some of their information and data and get some input beyond their specific family structure. And again, everything that the patient sees and including a more sophisticated provider report can be accessed by the clinician. The print is too small for you to see but there's a link on this page in Epic for them directly to go to the healthcare to the genome electronic and medical record. So we're obviously monitoring the use and impact of this as we roll it out into communities. We've already done the initial clinical trial which looked at the performance of our risk algorithms and recommendations versus an expert genetic team and a genetic counselor collecting information. So we had people for whom that was done in a very separate process and then we applied our health heritage and we demonstrated that we were as good as that process. So we think we have clinical validity. We're now going to test vigorously for clinical utility by our first implementation is at North Shore University Health Systems in Chicago, a very sophisticated system which has had Epic for a long time. But we hope then to go into other Epic sites. We hope to go beyond Epic. We're developing a relationship with Ancestry.com so that we can actually think about soon going into paper-based offices and extracting information from paper forms which as we all know, the majority of physician's offices in this country are still paper-based. So we think there's a great future for using EMRs as a source of information but not relying on them as the way to build decision support. Thank you. Questions? So going back to the family history information you indicated the sharing platform that you have. Are you reliant on the other relatives providing information or does the health heritage consume a more traditional patient-entered family history? The subset of that question is, does the health heritage support consumption of other tools that run on the current AH? Yeah, we link to the Surgeon General's, for example. And a person can enter a history themselves and the system has annotation where they did this information come from the EMR directly? Was it entered by the patient? Every family member has to give permission for their information to be transferred to their other family member. It's like Facebook. You have to friend someone to share information. It's the same concept. So yes, we are able to take information from different sources and the other thing that we've done is we allow patients to meta-tag data. We don't allow them to change any of the information in their medical record. But if they have a diagnosis or they see a problem that appears in their record by virtue of it's being transferred to us and they say, you know what? Yeah, they made that diagnosis a couple of years ago but nothing ever really happened with that and there can be this little meta-tag that sits beside that note and the clinician can decide to do what they want with that but at least when someone looks at it there's some annotation from the patient. We think that's also going to be a very positive way to help improve the quality of the data in these EMRs. I may have missed this or just need clarification. You said that once you develop a report with a risk assessment it goes directly to the patient and then the patient decides if they want it sent back to like a physician. Did they want it to go to their primary care physician, their OBGYN? Who did they want it to go to? So in the epic chart, did you just say like does some sort of notification go back at all? Yes. Okay. So notification will go back to the physician that this has been done and they're at high risk or does the physician have to go through the process? No. The physician gets a specific report that tells them this is what they are, this is the recommendations that need to be done. It's very direct. Only after the patient authorizes that to pass over to their provider. If they don't authorize it at all you've extracted the data from Epic. Nothing goes back to Epic. It sits in Epic but it sits in my chart. It sits in the personal health record of Epic. It doesn't sit in the clinicians component of Epic. The patient has to authorize that it go to that clinician. Is there like a legal liability issue from the physician's perspective? This is the patient's information. And one thing you have to realize is once the information is extracted from an EMR and it goes to an external application, it's your information. It's on your desktop. It's yours. You own it. You control it. So but if they have that information, it's from Epic and the physician has looked at Epic and not put it together themselves, are you opening up a door for a liability issue that the physician did not recognize? Like if time goes on and the patient does not say, I want this to go back. We hope that the communication, the two-way communication that sets up that the physician knows their patient is registered for health heritage. They know that they've done this. There's two-way communication that sets up there. But again, we think that as we move forward, the concept of patient control over their information is going to be increasingly important. Thank you. Okay, we're out of time.