 The floor is open now Yeah, I appreciate the The comments and everything I'm just still confused on the focus of Caesar With regards to electronic health records or emerge and whether it would be Important to not delineate because we want integration, but also not duplicate efforts in other programs So maybe I'll start You know, I've just recently started working with emerge and my understanding to date as is that emerge has been More focused on gathering phenotypes out of existing EMR data And I've done that fairly effectively in terms of trying to collect different phenotypes But I think as we've looked at particularly implementation of rare disease analysis The data that's within the EHR for certain disorders isn't well captured with existing data So I think one area that Caesar could really have contribution to is Developing better phenotype approaches that could be implemented within the EHR to better capture particularly rare phenotypes That I don't think are well captured today and figuring out how to structure that as well as figuring out ways in which We could develop clinical decision support tools that respond to the types of diseases and areas of work that Caesar is focused on That could be leveraged within the EHR environment that I don't think exists there today I'm sorry the chairs of the EHR working group are here, but the focus of the working group today has been Test reports that were issuing appear in the electronic health record What are the different places in which they appear with the format? And what are really the needs to improve reporting of genetic information particularly this level of genetic information into EHR So we've published two papers today and the last paper was really Survey of needs and we really specify very Specific specific issues around having test reports Be in data which can then be integrated into the kind of decision support tools It was just mentioned because currently most of the genome and exome data Appears in the HR and most of the sites Not in structured data, but as PDF and so one of our goals is to look at how labs could report structured data at a genome scale that could be used for the kind of decision support that we just talked about I Excuse me I Really resonated with both Heidi and Dan's Presentation and I think that the discussion that's followed it already is is really important from from my perspective It's important not to underestimate how primitive our ability currently is to take genetic data move it between institutions move it between laboratories and Integrate that with phenotypic data and what needs to happen in order to get good at that So that we can then build the clinical decision support in the broader infrastructure to support greater use of Genetics in in clinical care in the building and the building of evidence, so I think that I also think that the Volunteer efforts that have been mentioned the HL7 efforts the digitized efforts the GA4 GH efforts are great And I think I think it's amazing the way the community has come together to support them But from my perspective NHGRI is really in a unique position where It has I know limited resources, but funding for these types of efforts that in some way is unique and So as a recommendation looking at how NHGRI can potentially Increase funding for the types of standards that are needed to really robustly move these data So that we can combine them in the way that we have to as well as The the core implementations and the working with the vendor community to actually stand up the needed support That was Sandy Aronson who is leading both digitized and the eMERGE EHR working group My question is more for Dan what I think one of the areas where CESAR has made a very unique contribution has been in the legal regulatory analysis for genomic testing for Clea Among other things Including the FDA area and I wondered if you saw the CESAR New England Journal paper that Barbara Evans is the first author on it was a special report on making a database of Genomic data from clinical patients very similar to the database for adverse drug events and I'm sorry if you haven't seen it if you have seen it if you could reflect on if that's the sort of System that could then stand alone in the future that you're addressing. Well, so I think that you know and Any database that highlights outliers? Has some utility Right, but to the extent that we throw away 99% of The evidence that we generate about decisions made on behalf of every single individual by every single practitioner And we just choose as an industry to say well research is this kind of 1% activity and it'll it'll work its way back into it And and when we have the presumption that if we refine the inputs we present it in a way that is easy to understand on the screen That that's something good will happen the right thing will happen But we never actually checked to see what did happen then we have fundamentally a flawed model For healthcare becoming a learning industry. So that paper actually specifically Did not recommend capturing outliers But recommended capturing all genomic data on clinical patients and having longitudinal follow-up for outcomes downstream Sounds like a good paper could I get someone from Caesar who sort of interacts with patients on a Daily or weekly basis gale. I'm looking at you, but somebody else to to reflect on the differences between a Guideline that says do this when you have that and a patient who's sitting in front of you with a family and How do you implement those guidelines? Is there our guidelines hard and fast? Are there is there a role for for? Individualizing care beyond Even a guideline or a genotypically based guideline. Is that something that that we're still learning how to do? Is that something that? Is causing Dan Maces to have a little internal convulsion because because it seems to me that you know We we promulgate guidelines, but but at the end of the day, they are exactly that They're not hard and fast rules So and it seems to me that's the the the interaction between the the care system and the individual patient is where Caesar has made its Contribution to date and perhaps has its greatest potential going forward. Yeah, I think you're absolutely right And I suspect maybe Robert could help us with this conversation too that when you are sitting in front of a patient You can disagree with guidelines. You can disagree with the diagnostic criteria for disease You can disagree with the treatment plans that are guidelines for diseases. Many of them have actually very poor evidence bases One of the reasons we're so interested in the variant annotation area is because those guidelines are brand new And so people don't understand them the same way There are probably some unclear parts in those guidelines that can be fixed And so you certainly want to improve guidelines and get the best guidelines possible But where the guidelines fail the patients actually is an important thing that see sir can capture and should be focused on Well, and just to extend that so let's say you do believe that you're smarter than the guideline and you are and you do something Is it should should that decision not help extend and refine the guideline? Our resistance has been that guidelines should only be Implemented as decision support logic if they are perfect, right? But it turns out they only have to be useful and if you capture every single time someone looks at a guideline and does something else That's as valuable as as doing what the guideline says. So it's really quite a neutral mechanism It just depends upon the outcomes data. Yeah, and we've very specifically published our Responses to guidelines or the inconsistency of variant calls. These are all published exercises We've done educational sessions at the American College of Medical genetics meetings about where things don't You know don't fall where you expect them to fall in genomics practice to outreach to the community to let these lessons be public Yeah, I would build on both of those great statements I agree and say that the guidelines both precede and follow evidence generation and hopefully in an iterative process And I think that's one of the beauties of CSIR is that we have both taken guidelines and tested them And we have through our working groups tried to generate new guidelines and and that that is Well established as an iterative process that's continuing and needs to get to continue And I just want to add to that because this goes back to a question Bob Nussbaum Asked a little while ago. If we look at the guidelines that CSIR participants have been part of a lot of them Are the first step guidelines like how do we classify variants? How do we perform next-gen sequencing? It's a general framework for returning secondary findings But they're not necessarily the the practice guidelines for implementing that in different disease areas And you know Sharon mentioned the work in in the more cancer area But this has to happen across all the professional groups where that where they then develop guidance for how you know Which diseases are right for returning pathogenic variants? But if everybody's classifying variants wrong then no evidence is going to be effective later So we do have to start with these basic general guidelines for how to classify variants and get us doing the basic things correct If we're ever to build evidence later for how to implement this But I think in the next phase or CSIR I would love to see and I think CSIR can play a great role in helping facilitate those clinical guidelines within different disease areas So I'm I'm intrigued because I had written down rapid learning systems, which is similar to your Whatever you're calling them self-optimizing health care system, and I would challenge NHG and NHGRI to figure out if there are novel ways of funding this type of system To to actually do this learning process with genomics incorporated into it And the clinical decisions that are made because it's it's a very different process and not a traditional funding mechanism that I'm aware of anywhere being used and It's not necessarily something that every health system could do without some additional funding So I think it's it's an interesting approach to move away from this Sedentary sorry database, you know that just people can deposit information in if they want to as opposed to actually creating active learning for genomics Actually don't think the investment is very large at all. In fact, you know, it's really the mindset of saying that my job as a clinical decision support Rule generator or evidence committee is to say not only would I encode the phenotype recognition logic and the guidance to begin it I just think about well, how would I know whether the right thing occurred and Implement that as the secondary phenotype to just watch for somewhere downstream in the EHR the same way the event monitors Fired the rule in the first place So it really doesn't require either a dramatic restructuring of the technology or a lot more resources It just reminds me just requires the mindset of thinking about it as part of the package of designing this kind of infrastructure I agree and more the funding mechanism was to have it be publicly shared Rather than having it being done wonderfully in one system and not Not shared more broadly which which gets somewhat to Bob's point and the one that was being discussed before I made that point which is How how do we How does NHGRI and Caesar consider incorporating other efforts that are going on of? genome sequencing like the limit is understand your genome project or or Genomic sequencing that's not being done at Caesar sites and the learning that's coming from those sites because I feel like Now I'm an outsider from Caesar and maybe this is not something to say on the record But I feel very much like we're doing genomics, but we are outside Caesar And so therefore what what is being done at other sites is not being somehow Incorporated and learned from by Caesar Steve we have about five minutes left in this discussion sessions of people are burning Just raise your hand and we'll try to get to you Steve Steve Jaffee from you pen and from Caesar So I think just to reflect on this conversation and the point that Dan Mace has made a moment ago I think it's actually a bit more fundamental than that because the conceptualization here one approach Which is sort of like the Caesar approach is to fund a bunch of what you might call something like phase 3 4 clinical trials They are sort of single institution or a small collective in clinical trials of different approaches to genomic medicine and different clinical Context of quite different approach would be to say let's fund a learning health care infrastructure Which in which sort of pulls data as much as we can from clinical interactions observational data Which might actually be the foundation for prospective clinical trials But sort of layered on top of a learning health care infrastructure And I do think that there are some models to get to the point you made a moment ago that we can learn from although They're not perfect. I think the PCORI PCOR net Sites might be the kind of learning health care infrastructures that have some lessons for us The cancer cooperative groups are also the sorts of things that have some lessons for us So I think there's some infrastructures out there That might be some models that we could start to think about if this was a direction that the Institute in the community wanted to go I think in that regard if you just look back over the last 20 years the closest closed loop System is probably pediatric oncology Where the vast majority of patients go on a study which where there are downstream outcomes captured systematically It's good. It was a manual model, but it could be implemented with the same level With in a new domain it has a lot more features of complexity Katrina, I just wanted to follow up briefly on Debra's point and say that some of the Cesar Cross consortium activities are actually incorporating non Cesar sites So for instance on the study that we're doing looking at how different sites report carrier results Includes clinical labs that also report carrier results and looking at how those clinical labs may differ in their reporting from the Cesar sites But I also think there's an opportunity and possibly a recommendation could be made to have our meetings more open And include non Cesar Investigators who are doing similar kinds of work or a portion of our meetings To include non Cesar investigators so that we could have more of a two-way dialogue I was just going to say as somebody involved in the children's oncology group and having done Implementation of pharmacogenetics for four or five years that the most expensive part of those operations is Capturing the outcomes So I agree with you It's a great thing and that it also takes a change in mindset to get people to do it But it is by far the most difficult and expensive part of the operation To figure out what those outcomes are even if the outcomes are as simple as how was prescribing changed Did you really withhold coding for that patient? How long do you look in the future to decide whether you withheld coding for that patient if they're very manual? We have a terrific EHR that's fully penetrant, but it's very expensive manual process But so that As one as one implementer to another I mean I agree with that, but but The problem is that we just don't have systems We you know I don't I won't speak for you, but I know when we put our system in place We spend all the huge amount of time thinking about the clinical decision support What it would look like and having it word Smith by lawyers and pharmacists and informaticists But we did not spend you know a tenth of that time thinking about how we're going to capture downstream results So I think we're sort of suffering from that we recognize it and and that would certainly be something that Many people could work on Are we talked out?