 So, the final presentation this morning is a meeting report on the electronic health records meeting that Jeff Stroing is going to present. Great. Thank you. I think I'll be able to go relatively quickly on this. This was a meeting that emanated from the Arleigh House meeting on genomics and health IT systems, exploring the issues. You can see the planning group there. Really, Greg led this very ably. Greg, are you on the phone? I am. Can you hear me? Yes. We can hear you. So, you have a copy, I think, of these slides. I do, and I'm on the webcast as well. Okay. Great. Rex was there, and Terry was a participant in this, so they can speak up as well. So, this was meant to be a relatively high-level kind of meeting to look at the issues. It was a typical NIH kind of meeting with 90 participants or so, broad representation from academia, from numerous government agencies, very importantly from the Office of the Secretary, from the Center for Medicare and Medicaid, FDA, and the HHS Office of the National Coordinator for Health IT. We had some vendors there and insurers and professional societies and such. And it was meant to address these two kinds of questions or issues. What are those surrounding the intersection or some might fear collision of the health clinical IT systems in the U.S. and genomics and how we integrate the air of genomic medicine into this infrastructure? To look at what are the opportunities and challenges to fully realize the potential benefits of genomic integration and then look at what are the potential downsides and pitfalls. I'm just going to cover a few of the highlights. Greg is working on a first draft of a report. It was a very interesting, very good meeting. It was a very interactive audience and I think really had good discussions of a lot of the issues and I'm just going to highlight a couple of them. Importantly, one of the early speakers was someone from the Office of the National Coordinator. This is a group that got many billions of dollars in era funding to kind of jump start and sort of get the adoption of EHR going. He reported that a number of independent groups have come to relatively similar conclusions that by around 2019 they estimated there would be about 80% adoption of some form of electronic health record in the United States. That really both the health IT, the medical informatics EHR community and genomics communities are making progress. But they are largely on independent tracks at the moment and that the relationship between these two needed to be formalized so that they can articulate where they want to be and agree on how to get there and he continually made reference to the transcontinental railroad that we needed to decide where we're going to meet and have the same gauge railroads and so that when we get there it does hook up. To articulate a vision like this, to have a standards-based interoperable electronic health record system that supports health care and research. Another sort of very interesting concept that really sort of permeated I think everything was this notion of a learning health care system. I think there soon will be or maybe already is an institute of medicine report called the digital infrastructure for a learning health system. And this is the idea of a kind of continuous feedback loop of using clinical data and clinical decision support systems to both improve health care and do clinically oriented research. It was a very nice talk from Zach Kohane on this notion of EDGER or EHR driven genomics research. And I think something that caused both kind of excitement and concern was the blurring of the line between basic and health services research and clinical care that really, you know, using the EHR as a data source both brings all the tools we can bring very close to the patient. These are real life situations, but particularly from a human subjects and consent issue, you know, we really no longer have this clear separation between what's a research study, what's a research question, what's a research you're doing with the data versus what is a clinician doing or a clinical informatics support system. But this notion of having, you know, a kind of real time decision making and evaluating the outcomes of those decisions and integrating some discovery research was quite exciting. There was a lot of talk of standards and maybe a little bit harkening back to the Arleigh House meeting. I mean, there were a couple of camps, a few people sort of said essentially we need a new standard like we need a new hole in the head or like we need a hole in the head and a new one maybe, or that, you know, nobody can talk to anybody else because standards don't exist and until they are developed, we can't make any progress. I think sort of in general, the idea was that we probably don't need completely new standards, but the existing ones need to be sort of harmonized and be made more extensible. Mark Yandel gave a very nice presentation on the genetic variation format, which is very similar to the VCF format, but for example, the leader of the HL7 genomics interest group who was there had never heard of GVH. HL7 is the system for actually sort of transmitting data back and forth, lab values and things like that between systems. So it was more sort of the standard people getting together and understanding what each other's does and where they might need to be harmonized or extended was the focus. Certainly acknowledgment that the phenotype and sort of genotype standards are much less well-developed and that we need quality metrics for all of this data. This may not be the very best title for us live, but I put up here reinterpreting the genome. I mean, following up on that last point, there's a lot of work that needs to go into developing how we would interpret a genome variation, genome variant or a whole set of genome variants today, but as difficult as that is to interpret the first time, we're going to need to continually reinterpret this for many years as our knowledge of the clinical implication of variants comes into being. And this was pointed out particularly by I think some of the vendors that the electronic health record is a medical legal document that it needs to both document what the decision was made, but all the information that was available to that clinician at the time. And if they're making decisions based on a set of genomic variants, you know, how do we sort of harden what information we had then because next month there will be more data that might help you reinterpret or further interpret that genomic sequence. The whole genomic sequence or maybe not even the whole GBF needs to be sort of part of completely within the EHR, but how we expose various parts of the GBH file or the genetic variants was discussed and needs to be worked on a lot. And also this notion of sort of critical values of a genetic variant file. Even if the clinician or the system doesn't support access to, they haven't asked for results on a given sequence variant for a gene. What if based on whole genome, whole axome sequencing, you know that this person has a variant in a very high-penetration gene. Is that something that needs to be reported? But this is, these are, I mean, many of these are issues that have been encountered in other areas and I would sort of, there was some talk, you know, whether we're getting into the genetic exceptionalism again because from an EHR vendor they said on one level a genomics test is no different than any other laboratory report. It gets ordered, you get a report and the clinician interprets it. But and they face this in laboratory testing when a clinician sends off a blood tube for a serum panel and they ask for electrolytes. These machines, they do all 500 tests and they just don't, they only report the five that the person requested and paid for. But they do have this notion of critical values. If something that the clinician didn't order is so far out of range, that gets reported. There may be some, you know, are there genetic variants, such genetic variants, there's sort of, you know, a lot of issues like that. I think I just have two more slides. The research was mentioned many times as kind of a trail blazer. There were a number of talks from emerging investigators which went off very well. And it was really, I think, called out as an example of the kind of groundwork that's necessary to figure out how to go forward and that sort of continued expansion and extension of this model was really a good way to go. Phoenix as well was called out as a way to begin to help the more on one level thorny problem of standardizing how we just defined phenotypes. We've talked a little bit about data security and data storage. Actually, a somewhat surprising, I think, conclusion at this was that since the medical record would be looking at maybe a GVF level or even a part of that file, that sort of data storage wasn't really seen as being a big issue. Because they wouldn't ever consider the kind of reads or lower-level data. In the interest of time, I think I'll just sort of touch on these very, very briefly. We haven't coalesced around a complete set of action items yet, but these were some of them that we do. There is a need for some leadership in the standard space in getting these two communities to talk together. There was a fair bit of talk and need a lot of work on sort of really sort of maybe moving beyond the standard informed consent model of an individual giving consent for a particular study. That if we think of getting the best use of the electronic health record and the integrated genomics information that you need ways to mine this effectively to support clinical decision-making without having certain records and certain parts fall in and out, we need to sort of get to ways of doing that. There was certainly agreement that we need some mechanism to come up with some sort of set of standards in a way to sort of come up with what are the clinically relevant variants and how do we make that accessible to clinicians in an understandable way. That we need pilot programs for demonstrating the clinical utility of genomics and EHR integration that we're just showing that we can integrate them. And isn't this neat and cool? We need to show that they actually provide clinical benefit and are cost effective as well. I think there was some more, but I think I'll stop there and let Greg and Rex and others make comments. Thank you. It was a really nice summer of the meeting. It was really a great meeting. One of the things that struck me was the breadth of folks at the meeting ranging from electronic health records vendors all the way through genomics types. And just to emphasize a couple of points that Jeff made. In terms of standards, there's a strong sense that as the electronic health records initiative rolls out more in the Office of the National Coordinator implements the rules for meaningful use. And as the health care community implements meaningful use and deploys EHRs, it's going to really improve the quality of the data in electronic health records for future research. So it's a great, great opportunity and something we really need to pay attention to. And I think looking at that in light of the kinds of things that Emerge has already been able to do, it's already pretty good. But if it gets really good or excellent going forward, it's really something we need to pay attention to. One of the things that I think strikes me as we think about these standards and how do you actually get a clinician to implement them. We need to think hard about how we annotate, and Jeff alluded to this, but just to put a final point on it. How do we annotate the genome and the genetic variants that are actually clinically relevant and clinically actionable? And we need to really be paying attention to that. Because as we go forward, that's going to be a key piece of this. And sort of in my own fantasy, I imagine it would be really cool if there was a centralized data feed, a GenBank kind of like thing that had these annotated clinical variants and some quality score, if you will, that's attached to them, that any hospital electronic health record could simply point their system at and get a web dump or a web services feed of this information and be constantly up to date. The problem, of course, is there are going to be local clinical standards that need to be applied and variation across different sites to implement that. And as I was talking about this at the meeting, I saw people like Peter Good just blanched as they thought about what the responsibilities and costs of doing something like that will be. So I think that's going to be a real challenge going forward. And then I thought, actually, I was very impressed by Fen Genie, and I wondered if that might actually be a step in the direction of that kind of a data feed. So we should pay attention to that. And then the final point that I'll make is I thought there was also a fair amount of discussion of some of the LC issues about, really, what's different here is that there's so much data. And it affects the families in ways that maybe some of the other things don't affect, some of the other kinds of measures we're used to dealing with clinically don't affect families more broadly. So that was an important issue that needs to be thought about. This idea of what level of data do you share? Some people at the meeting were very clear that they never wanted to know their APOE alleles because they didn't want to have to deal with that consequence. And I think the other point that Jeff made that I would really emphasize is why is this any different from anything else that's in a clinical record in terms of lab value? We all know that your total cholesterol level is probably a pretty good predictor of cardiovascular risk. And so maybe we're making too much to do about genomic information. Lucia, or Neville, like Greg, say anything you think. I just wanted to make a comment about, Rex's comment about Fijini. So working with NCBI, we've gotten to hear a little bit about what their resources are like and what they're thinking about. And I think they are also interested in clinical annotation. However, the data that they have right now are data that are essentially submitted by outside users as clinically relevant. I don't think there's a common standard. And I think that's really what's missing, although perhaps having some annotation at all is better than nothing. I think the standards issue really needs to be addressed. Yeah. I was only there for the first day of the meeting, but I think it was a good opportunity to have sort of those at the federal level who are kind of putting together all the electronic health records and the mandates for those with the researchers. Because they really seem to be a disconnect between recreating this huge thing that will be good for clinical care and research without a lot of thought into do I want my stuff being used for research, as well as researchers putting research level information into the electronic medical records and kind of blurring different types of information. The second thing I thought was very intriguing, there was a lot of talk about informed consent. And everybody immediately their eyes glaze over and think about the 40 page consent form, which we make with regularity. And one guy got up, I forget what his name was, and really suggested. Perhaps we should look at this as public health reporting. You know, you get a gunshot wound, you're going to be reported to the police. If you have high cholesterol, perhaps this should go into almost a public health reporting mechanism. So I think there's a lot of, there's some excitement about some different ideas, be it notification, more than informed consent, or you're looking at different approaches. Greg? Do you have any? Hi, yes. It's Greg. I don't have much to add to those comments, except to say that there was some debate, I think, going into this meaning about the timeliness of beginning to look at this intersection. And at least for me, and I think for some of the other attendees, there's a fairly good consensus that due to the huge inertia and amount of investment that's going into the public effort to have EHRs be adopted nationwide, that it's better to be thinking about these topics and working on them now, rather than trying to retroactively get a huge number of vendors and academic systems to think about incorporating genomic information 20 or 30 years from now. Great. Howard's pointing up again. I wasn't sure you were talking about lunch upstairs. No, I have a comment. I am interested in that, as you can tell. But that's a different story. I think that a lot of the comments that were made here about trying to pull together some consensus around which variants are actionable. Also come back to the presentation we had on the microbiome, which is there's a lot of inertia out there. And I don't know if you're starting to see this, but we're starting to get clinical ads where Chinese companies will do your microbiome for you. And they should be calling it crap in a box, because that's basically what they're asking you to do. You're crap in a box, send it overseas, and there you go. But you can't send Chinese crap over to the US. So that's a different story. But that idea of annotating these things, I don't know if it's an intract issue or an NCR issue, but it's coming way faster than us so-called geniuses on the advisory council realize. And I think you guys need to act on it with or without our help. It seems to me, it's a tough issue in Eric's slide about the CDC's population health genomics office going away in this EGAP program. I mean, that was incredibly labor intensive, went through a relatively small number of genetic variants. Yeah, don't do that. Right, I mean, so we need something in between that and a completely computer-driven thing that simply spits them out. There has to be some level of human curation, but it can't be at the level of EGAP, and it needs to go- Many of us here are having to do it for our own institutions. And it would be nice if there was done at a national level or a regional level, not each hospital. I think I've nearly gotten this back on time. Yeah, the speakers this morning did very well keeping us on time. I think we have time to go upstairs, break for lunch. We can be back by one, let's say 120 so that we can really start at 130. And if all the presenters in the afternoon can do as well as the presenters in the morning did in keeping on time, I think we'll be in good shape. Yes. So you're at 120.