 with the same microphone. So again, I would thank the whole members of the group. We did exchange slides almost a week ago, which was, again, a violated all academic rule. So the next slide. So what I'm going to do is sort of summarize some of the recurring issues that are the next slide. On, you know, that we've heard many, many times already today and then make some suggestions of things that perhaps I could be, could be future direction. So we've talked a lot about the research versus implementation and eventually clinical application. And I think there's also a distinction in there between the GWAS studies and sequencing. GWAS is essentially a platform developed for gene localization and gene identification. And I think there's, you know, there is certainly some returnable information in GWAS. But most of us, I think, have the bias that most of that is not intrinsically returnable, not paternity or some shifts are designed to test certain polymorphisms. There's also the whole issue of incidental findings and how you deal with those in research versus in clinic. You're doing clinical de-sequence, and I think it's a much retractable problem. And then there's both inpatient preference. And I would extend what we've talked about some from just the consent of the patient to all the way to the family and the 2D population. There's been some talks about the Israeli application. We certainly were discussing last night the population screening and the pediatric. At what point is a population, might we say, yeah, this is a good thing for the population to know, especially if we're paying for everybody else's health care. CLEA-approved lab results is all of the bug there, and you need to address that up front, I think, and we'll talk about these again. And then how to interpret this in the EMR. So in the next slide, I think the question is, is there ever a GWAS study that you would want to return? And look them in the model of 23andMe from the VA to actually use 23andMe as our platform at one point. And I'm very glad we chose not to do that. It's an actually interesting model, and we used it in a class I taught a year ago. And so I've had the opportunity to do it. Most of the faculty found themselves playing on the website for well over five hours. I didn't. And so might one use return of GWAS results as an education tool? And could you even engage a population by doing that? Is that wrong? And paternalistically, I'll say, oh, there's nothing there that's actionable. But people love it. And could we even return it on a group level? I mean, certainly we publish things and say, yeah, this population has this or has that. I think in the GWAS now, there's cardiac risk profiles that are coming out and some other things that are distilled down. But could this be used? And might people change over time? I think, again, back to the 23andMe model, they are Google, right? And on the back end, they're tracking exactly how much time people spend in different places. And I think we might be able to refine the way that we'd interact with the population over time by looking at those data. Next is sequencing studies. You know, I think there's a misperception of providers that hold genome sequencing gets you the whole genome sequence. And it really doesn't. There's a lot of details. Why would they think that? And so there's obviously some education. And there's still, for the rare variants, there's much firmer disease associations. But there's a lot of problems. And then there's a CLIA issue that we've talked about before. And again, I think to the extent that we can do tests like that in the CLIA lab, Mark just mentioned that this is durable information and we could use it forever. And then my guess is that the list of 56 genes will be expanding over time. And we may have institutional differences between those that who curates and how do we do it. And on the next slide, I think I talk about consent. We've already talked about that a lot, especially in the last one. But I work in the VA. And there is the requirement that within the VA we get documented consent for any DNA-based tests. In contrast, if I get a test of a drug level, it might tell me the same information or a test of an enzyme level, that can be covered under the general treatment consent to treat. And not a specific consent. Of course, you're supposed to discuss your care plan and the outcomes of all tests with a patient. But I think there's a whole spectrum of genetic tests from the simple pharmacogenomics to Huntington's disease. And of course, there needs to be a consent process, not just a form sign for Huntington's disease or whole genome sequencing. But at what point are we going to say, yeah, running the pharmacogenomic panel and incorporating that in the electronic medical record, that's not research. That's not really any different than getting a creatinine before you prescribe the drug, in which you wouldn't get specific consents or any clinic. In the contents of research, what, five years ago, only 10% of research consent documents mentioned return of information. Now, I anticipated it changed a lot in the last year. But some only state that subjects will not be informed. At our large trial in the VA, the Million Veteran Program, the consent document clearly states that no results will be returned. But I've had the opportunity to discuss with many of the subjects, and they would like it to be returned, very much so. And there's a, I think, how to, that's a very strong preference. But most research is not done in clear certified labs. And so there is always a need to retest, often a need to redraw. And that's an impediment. Many times, you have to go back to the IRB. And as we're doing an increasing number of tests with direct clinical relevance, it's just like doing an MRI or a CT scan. In general, that would be done in the routine hospitals in the lab. So the next slide is patient preference and how we can actually address what people want. Kathy Hudson, this was prior to the MVP, found that 96% of the veterans would want their research studies released. Now that's, as a general rule, obviously that might change when it comes down to a specific issue. And we've got to honor that 4% or 10% that doesn't. But I think that the population, the patients and subjects and participants all across the board want that. We have often tried to have an a la carte menu. We were talking about last night again the idea, oh, I want this result, but I don't want that result. And the only place that I see that in medicine really is in cardiac resuscitation. I want an ET tube, but I don't want to be shocked. I want drugs, but I don't want, you know. And if you're being a practicing internist, you would never say, well, I want to know about my lungs, but don't tell me about my kidneys. That's not really an option. So how do we get to the point where this just becomes routine care? And I think that Merge can help with that. And then do we return these results to the provider, to the EMR, to the subject, or to others? And so I think this is all a good opportunity. So the last slide here, I hope, is just the next slide. I think we are tending towards leaning towards that factor of fear. So the next slide. Oh, actually, the last slide here. Computerized decision support, I think, is necessary to return results. Most people are not prepared to accept the results in their clinic. It's one thing, so anyone at this table think about a cytochrome polymorphism. We've been thinking about it for a decade, whereas in a busy primary care clinic, you're not going to find that. And so one of the areas that I think we can look at is the way that clinical decision support is implemented. So often it's focused on drug interaction and it's focused on just something wrong. And maybe you've just spent two minutes doing something wrong, as opposed to using human factors engineering to push that information forward in the encounter. So how to deliver that is another area where we can go on. And so the last slide now is just the opportunities. I think how immersed can be is for education, how population preferences change over time, especially with targeted education after results are returned, and then looking at that human factors engineering and engaging the EMR in CDF and pharmacogenomics is an obvious good place where the implementation in the different systems across the e-merge network could be used as a tool. And I'm sorry I ran over.