 So, yes, thanks. Casey Overby-Taylor, I'm at Johns Hopkins University, and I've participated in several of the NHGRI projects, including Co-Chair of the EHR Integration Work Group of eMERGE, and I previously participated in the IGNITE SIG group, which is now the Clinical Informatics Working Group, and contributed to the Clinician EHR Work Group, so I'm going to talk about some of the knowledge management tools associated with those projects, and then also, given the session yesterday that focused on implementation science, I changed some of my slides so that I can go through some of the frameworks within the clinical decision support literature that might be relevant to kind of augment some of the models that are being discussed in this venue. So just to start this out, I like to show this image to just give an overview of the vision for clinical decision support as a way of bridging what's in personalized medicine into realization, and as you heard earlier yesterday in terms of innovations being able to be seen in practice, they take 17-plus years, and so this is one of the approaches that could help to address this. This is a paper that reviewed the literature from CDS on genetically-guided personalized medicine that identified 38 articles and concluded that to maximize the clinical benefits from ongoing discoveries in genetics and genomics, more research is needed in this area to identify the best approaches, and so as you know, this is fitting very well with this idea of implementation science as well, and how do you implement CDS the best way in order to address these barriers. Some of the barriers that were identified in the literature and discussed included limitations to genetic proficiency of clinicians, limited availability of genetic experts, and growth of genetic knowledge bases. And so this is just an overview of the topics I'm going to go through. So first, outlining some of the challenges for genomic clinical decision support and talking about the implementation science literature and how it's related to genomic decision support, I'm going to focus on some of the goals for eMERGE 3 in terms of genomic clinical decision support implementation, then I'm going to shift a little bit to talk more about managing shared knowledge for genomic clinical decision support specifically and the tools to enable knowledge management with a focus on NHGRI funded projects. So this paper was published in 2013 and came out of eMERGE where we highlighted some of the main challenges to implementing genomic clinical decision support and the main points that were brought up in this paper are managing shared knowledge, improving effectiveness and establishing decisions for architecture and standard approaches. In terms of managing shared knowledge, given the focus of this talk, I just wanted to highlight some of the papers that inform these challenges that were highlighted. So first, knowledge management solutions often are not accepted without customization and that's something that's been found in clinical decision support literature broadly. So this publication here highlighted that for several clinical decision support demonstration projects and the same is true for genomic clinical decision support. This paper with several of the authors in the room highlighted that there's often a reliance on expert communities and so that's part of the customization activity that occurs is getting input from expert communities. In terms of improving the effectiveness of genomic clinical decision support, one of the main challenges is the lack of institutional and clinical acceptance of supporting evidence and one way to kind of approach this challenge is to consider user interface characteristics, information content, integration of workflow and decision-making processes and while these are things that we suspect would improve the acceptance of genomic clinical decision support, there's still work that's needed to really understand how those features translate into the acceptance of genomic clinical decision support and so that highlights again the need for implementation science approaches and informatics approaches. The third point that was brought up is around decision support architecture and standard approaches for genomic clinical decision support and one of the main challenges that there's a lot of variation in the architecture and standards are very much needed to scale. I'm not going to talk a lot about standards because that's going to be discussed in following sessions in a lot of detail. The main points that people generally bring up in terms of limitations to using standards are that there's too many to choose from and it constrains, once you choose a standard it can constrain what the user can encode and so you're limited potentially in your scope so often standards for genomics may need to be refined to incorporate the scope that you're interested in. So within implementation science, one of the key points I took away yesterday was identifying the what and I really like the Peter Provenos example where it was the process of creating a checklist that actually improved outcomes and so there's a lot of effort that goes into that defining the what in terms of clinical decision support and for genomic decision support specifications the what is often defined within the context of IT capabilities and so it's clear in the literature that there's insufficient decision support technology often and so this can require additional IT development and resources which may or may not have been in the original scope of a project and so when you're assessing capabilities we'd love to often design a CDS tool that's separate from the EHR because maybe that's developed a lot faster but if you want people to use it needs to be integrated in some of these the capabilities and the internal builds have to happen in order for that to happen. There are the good news is that there are some non-technical solutions for decision support that can be used so one of the points that came up yesterday for example was a consultation service so things like that can be non-technical and then often in these studies and in implementation that starts as a research protocol you initially may have a study team to support some of the processes that can later be translated transferred into technical solutions and so there are some frameworks that exist from the clinical decision support literature to assess implementation challenges and to guide local approaches to implementation. These are a couple that we described in the paper and you see that there's actually very familiar concepts that are were discussed in some of the implementation science models but I think one of the the one of the key things is that there is this focus on the tech technology and software and hardware and aspect and in the paper we also define this framework given that some of the existing frameworks that we that we assessed didn't really didn't address specifically what we were looking for and so when we're defining the what genomic decision support do we need we can consider kind of three areas the stakeholders the transactions in the in the clinical systems and these can be distilled out to questions around around this I think that the key thing was characterizing what are common genomic testing processes that clinical decision support could help enable and this scope is it has a health care provider or consumer and the consumer as to two of the main stakeholders but there could be other stakeholders that are not actually represented in this but for for our goals this this seemed to fit the examples that we were exploring and so just to give a couple of examples here first just going going through the overview of e-merge three when we apply the framework it's pretty straightforward for what we were doing in e-merge where we're focusing on the end of this pipeline here so this is very high level there's a much more detailed version of all the steps that went into this but the the current process within the e-merge three is to screen patients for genomic sequencing and so this involves the preclinical recruitment which happens at the e-merge sites and then this the sample collection and sending to a centralized one of the two centralized sequencing labs both the VCF and raw data are retrieved and stored within the repository and the report is delivered to the sites which also has managing interpretations of the raw data associated with those reports so where we are now within the at least the EHR integration work group where as we're seeing how we can leverage the EHR for processes for returning the results in terms of summarizing outcomes that's there's a outcome work group that's doing much of that aspect but when we're just thinking about using the EHR for return of results when we consider this framework once more this is again kind of like a high level simplification but when you consider what are the relevant transactions you're retrieving genetics and genomic test results so that's one of these transactions as listed here when should the activity occur so post-analytic so after the genomic lab ritual sends a report and how should it be initiated by who and it's a health care provider who's receiving that report and where should the data be pushed or pulled from it could be the EHR but there could be other mechanisms so this is kind of a discussion piece potentially another area if we were to consider other upstream point for genomic decision support is for patient screening and so in that case the relevant transactions might be reported personal data family history and pedigree data and that would happen prior to genetic testing and that could be initiated by health care's consumer and the PHR could help enable that and so when you drill down even further after that very high level you can think of like what is the what is the CDS content and documentation templates for data collection is something that you'd have to build within the EHR and when that occurs you have to define the setting so maybe the outpatient setting what's the workflow context it could potentially happen between visits and by who this could be the patient and where should the data be pushed and pulled from this could be internal off-shelf functionality of the EHR and it could be active clinical decision support although it's the clinical it's the CDS features that should be would be different in this context context of active decision support so in this context the trigger is time so so maybe 24 hours after a patient visit and the the input data element might be that they've they've had a visit a patient visit and the intervention in this case could be an email that says can you fill out this this documentation of your of your personal data and the choice in in this case is to either ignore that billing that out or it could be actually completing it at that time so these these are features of decision support that are included in taxonomies that have been proposed by others involved in clinical decision support so I'm going to move forward because we're getting low on time so so I'm shifting a little bit just in the to overview of managing shared decision support and I'm going to go through very briefly the tools that are used for this so this is an overview of kind of general steps for for genomic clinical decision support in terms of how you would envision this within sort of a learning health system view so you would build and revise this genomic decision support based upon clinical practice guidelines often you can't there isn't there may not be a guideline specifically for what you're trying to do and so this is this can be local policies or discussions around what is included in genomic clinical decision support so I included that as a bullet in and once it's built then you we would want to publish the genomic clinical decision supports but particularly in networks such as e-merge and others and so the computable genomic clinical decision support is often captured locally and in an e-merge in e-merge case the clinical labs also have structured interpretation that they're capturing as genomic decision support and so this is kind of the main area where there existing tools that can help address this after something's been published then it's being used and you'll want to monitor the decision support so those are this this process has been pretty much covered with among the group yesterday so in terms of building a revising decision support there there's the spark toolkit that was brought up yesterday and I that helps to provide guidance on the implementation project process and so I'm just pointing that out because it was discussed yesterday I'm not going to through it in detail but I think that's one of those is a tool that could be very useful there and then better engaging stakeholders I do think that this is an opportunity for new tool development and and there's not a lot in this space this space currently in terms of publishing genome decision support briefly these are three or these are four tools that have been discussed within these groups so I will the slides are shared but I'll just point out the publication so the genomic CDS sandbox is something that was brought up at a previous genomic medicine meeting that was focusing on genomic clinical decision support and the idea is to have a sandbox to that's made available with pre-configured genomic genomic tools and clinical decision support so that you can try things out in the sandbox environment before implementation and that's something that is envisioned but has not yet been developed also within ClinGen they created a HL7 compliant search interface for existing resources and provided guidance on how to how to revise those resources so they're compliant there's another tool that was developed within the e-merge network by Luke Rasmussen where it's called docuBuild and we found that there are several institutions that were that were developing similar content and had the need to brand them in different ways or include local content such as reimbursement rules for their institution but the core text could be shared and potentially branded for their local needs so it's just a tool to help with that and CDSKB is a large effort that includes both community engagement as well as a collective library of artifacts that can be contributed by community members who are involved in clinical decisions support efforts and so the focus today has been on EHR integration clinical decision support and technical implementation and one of the really useful aspects of this site is the archived webinars where community members can present and everybody around the country can can learn from their implementation efforts and and ask questions and it can it builds discussion on topics so that we aren't reinventing the wheel there and then also there's a current effort serving sites about genomic medicine about the genetic medicine data pipe pipeline and so that's another contribution where you're they're getting community input so I'm going to skip over these last ones but the summary of this presentation is really we can learn from efforts in the broader CDS community to help address challenges for genomic clinical decision support and implementation science models can be complemented by existing frameworks to guide understanding challenges and approaches to implementation we also should consider further investment and plan and underdevelopment tools such as the four that we presented I presented here and the design of tools can be extended to support precision medicine and there's a whole section on precision medicine so I skipped over some of those slides but that's my talk thank you thank you Casey any clarifying questions remember we'll have the bigger discussion in a little bit all right in the absence of that well Larry Bab from the broadest who will be up next talking about some of the bi-directional data flow that he's been pioneering over the years