So, there are five steps to design thinking, from empathizing, defining, ideating, prototyping, and testing. I've added a sixth step. So, we've built a tool that is been shown to benefit patients in their decision-making with low-grade carcinomas, but there's another step that we have to work on to really bring it into clinical care, and that's our deployment phase. That is really because our care providers are incredibly busy each day, and there needs to be a way for this be integrated into their workflow. If a tool is not integrated into the clinical workflow, it may as well not exist today. Our physicians and our nurses are doing a tremendous amount of work, and they're spending several hours documenting to an EMR. They generally don't have time to go look for an external tool. It needs to be part of their clinical workflow as they're visiting the patient, as opposed to something that they would have to go out and search for. So, one decisions are how do I need to build the solution, or can I just integrate it in an existing electronic medical record? We are seeing a lot advances with electronic medical records. They perform these key functions of maintaining a directory of the patients within that health system, being able to create, collect, and maintain records on those patients. So, they collect a lot of records that are not generated within the EMR that might be coming from a laboratory or from a radiology reports. They provide care providers a central place to manage problem lists for those patients for being able to, through an EMR, being able to place orders for medication tests and procedures, as well as manage clinical documents and notes, like discharge notes and clinical summaries. It also be able to be their workflow management tool for care plans for patients who are in the ICU or need to have a sequence of steps being done as part of their care. Electronic medical records is also a platform, really has a directory of all the patients and the care providers. So, it should be seen as a hosting environment or an application environment for third-party clinical decision support solutions. So, I think one of the first questions you should be asking is whether you should be building it, buying it, or can you actually just implement your solution directly with your vendor, with your electronic medical record through configuration? We're seeing continuous innovation from our EMR vendors and their capabilities. I put together this rough guide of where I feel the functionality currently lies. I see this as, for most calculators, a lot of them now can be built directly in the EMR interfaces. It makes it a lot less data transcription for the providers to be able to just build it as part of configuration with the EMR. Some very complicated calculators might need to be built as an external solution. Condition rules also are, there's a lot of great work flanges now, as part of electronic medical records, in terms of implementing simple if-then statements for follow-up and reminders, and for alerting conditions. As we start moving into data aggregators, they can start showing sparklines and different graphs of trends of different data within the EMRs. They're making a lot of advancements, but they're also, when you're looking at specialized views, sometimes you're going to have to build a separate solution to be able to integrate into clinical care, like we did for the PSA levels. Then when it comes to information retrieval tools, within the EMR, there's usually the ability to search within a patient's record based upon the text documents in the reports, but generally, it's not across other patients or connected to other knowledge bases like PubMed, or clinicaltrials.gov. Then other heuristic models, which are really knowledge bases that are generated by these medical societies, that can plug into an EMR, and lastly, probability models like what we've built for prostate active surveillance are still highly specialized and are more likely at this time to be built externally to EMR systems than internally. However, EMRs are innovating rapidly, and we're seeing a lot of improvements in usability and workflow. So, this is something that you should be constantly evaluated to see whether you can build, or buy, or integrate. You should always look for what's the fastest pathway for developing these solutions and implementing them within your care contexts. Obviously, if you can implement it through configuration, it's going to be a lot faster than having to buy a solution and even, ultimately, having to create one on your own. So, I think it's an exciting time for decision support tools. If you think of the EMR as a platform for third-party applications, Dr. Green presents a really nice model for what are some of the things that need to be done to integrate these decision support tools into clinical care so that they are seamlessly integrated, and they can be used effectively by care providers. So, on the bottom here is electronic medical record. It needs to be able to do three main functions and that is: provide data within the EMR, the information model needed for decision support tool; it needs to be context integrated to be able to launch the decision support tool and be able to bring in user interfaces into the EMR for the specialized case;as well as be able to receive and collect external documents like reports and to be able to be able to then share them as part of the generalized medical record. From the decision support tool, there needs to be able to take this off, and this is one of the hardest parts is taking data from the EMR and mapping it to the decision support tool. There's several standards bodies out there. The Health Level-7 has an emerging standard called The Fast Healthcare Interoperability and resource standard called FHIR, which is a way of describing resources in a way to be able to map them into a decision support tool. The values can be mapped to terminologies such as SNOMED, where then you can link it to an inference again. So, let me give you an example for our prostate tool some of the steps that we had to do. So, we had to extract PSA lab values. We had to pull the pathology report to be able to pull the Gleason scores out of them. For the MRI, we had to look at the radiology reports to pull the Py-Rads scores, which is basically their estimate of how suspicious that tumors are within the prostate. So, it's really important not to have misconceptions around that data. That data needs to be very well clearly defined. The more you map it to a standard like a terminology, the less likely you are for misinterpreting that data. So, being able to make sure that that's tied to a terminology is really critical. Then the decision support tool needs to provide an inferencing engine. In this case, we're using Bayes, which is a probability mapping, but there's really many machine learning algorithms out there today that are being implemented such as support vector machines, deep convolutional neural networks, and regression keys tied to random forests. So, there's many different ways of doing inferencing engines, but then there needs to be a model, a knowledge base that uses that inference engine that takes that data and creates a prediction and helps you inform a user interface to then present back to the user. So, ways of providing contexts integration, there are some emerging standards. One's called CDS hooks for being able to pass a launching information and context back into the EMR. Then when you provide that user interface to the care provider, and they've decided upon a care plan, you want to be able to generate a report of what the decisions are and bring them back into electronic medical record. That needs to be able to be stored into the EMR and searchable by other care providers to see what was done at that clinic visit to be able to help them make their decision. So, this whole process requires a good deal of integration and needs to be able to be tied into the clinical workflow. So, when they're in the clinic visit, they're in the patient's medical record, they can just be able to click on a button and launch this tool to be able to see and to be able to provide that to the patient to be able to help them and inform them with their decision-makings. On the patient side, there's also a patient portals where the patient can directly get access to tools where if they wanted to follow-up for patient education and be able to interpret the results or to be able to help better understand the results at home. Great. Thank you.