 Welcome to Public Good At-House. My name is Aretha Simons from TechSoup. Today we have some amazing people who are going to share some creative tech solutions to address COVID-19 and other health demands in the community. Here we have today Roberto Valdezad. He is the founder of Bystock and Cillibrand LLC. He's gonna share an app with us that helps doctors diagnose and treat patients promptly while engaging with SMS tools. My name is Roberto and I'm the founder of Cillibrand, which is your organization continuing the development of Bystock, that is a patient engagement tool to diagnose and treat people with disease promptly. We actually started out this project last year with the start of the pandemic through the MIT Hackathon. Claire Danotte, who's a PhD in statistics and teaches at Chicago now was a lead data scientist that helped with the algorithm. Freddie Bamburri, who's a PhD from Cambridge in microbiology was the one who provided a lot of the medical assistance. And then Anurag, Jesse and Tal and myself had been part of the team that has been developing this tool. We're able to take this tool and make it available for more things than just COVID as we have already done it for COVID and it's already available online and can be used by anyone. We were able to increase the accuracy of lateral flow essays. So that's a rapid test for COVID from a baseline of 70% to 16%. By using big data that was provided by England's NHS and other public available data during the pandemic last year. All this data is unidentifiable. So there's no risk that we manage data that was private but it was of great use as we have been able to deploy it and be used for multiple countries. We're actually in the process right now of adding data for India to help with the current crisis underlying. Why do we think this is a great tool to use not just for COVID? Of course, with COVID as a new disease as a novel virus, it was hard to diagnose it at first. A lot of the people would show up into clinics thinking that they have symptoms or even having a test that had a false positive. But once they got there they would realize that they don't have the virus but potentially might get infected by others who actually do have it. The opposite was also true. People that have gotten multiple tests that come out of negative actually have plenty of symptoms and a case that would correlate with someone that has COVID. Therefore, they should be treated promptly and if they're treated well then the disease should be well managed. For this and other diseases 25% of patients are misdiagnosed the first time. For some uncommon diseases this might take five or more years to get diagnosed. And it's really unbelievable that it's still happening in the 21st century as many of these diseases can be correlated to specific medical data such as well even now that we have the Human Genome Project to more than 80% of diseases even for uncommon diseases. The way we are presenting this solution at this time is a patient engagement tool in the form of a web app in which people can fill out the survey and answer questions that are pertinent to the case information without giving any private information. And then this will be fed to an AI they use all the big data that I talked to you about to correlate that with other medical cases and then provide a result that they may have and with a quantifiable confidence of that. We are working right now in getting the suggestions for not only providing them with a quantifiable idea of what they have but also who can help them with it. As you can see in this demo, this is a web app that's already available and can be used by anyone who's wondering if they may or may not have COVID and the one that will be shared with India. Through here, people can actually click to watch the video and make sure they're taking the lateral flow assay or answer the questionnaire to start feeding the information to our AI. This is available in English, Russian, Spanish and Portuguese at this time. So when you click the video, you'll be able to see a video that was provided by Abingdon Health in which they explain how to take an ABC19 rapid test. Through this, people can make sure that they're following the procedure correctly. Our survey was designed by UK medical doctors that were actually connected through the NHS. That's why we were able to use that data on identifiable data. So it gathers all the important information without sintering with any private data. As you can see with the user-friendly survey, all the important patient case information such as the exposure risk, depending on where they're at, where they have traveled, if they're vaccinated or if they are also among populations at risk are taken into account to provide a final diagnosis. All this information is crunched by our AI but we are not storing any of the identifiable data. So although we have all this results stored, we wouldn't, and no one would be able to really trace it back to anyone, which makes the tool KIPA and GDPR compliant. So as you can see here, this is an example of the result that people would get. Even if they had a rapid flow of say that came out as positive, but they have been taking care of themselves and they haven't been exposed to the virus, then this tool would actually provide them with a confidence interval of the unlikelihood of that result being correct. But we would still recommend them to get checked with a practitioner if they get onset of symptoms. The same would still apply for the other type of patients that may not, that may have a negative test result, but they still feel the unwell and their medical case actually correlates to someone that may have COVID. So this type of information is visually understandable to people and provides some sort of confidence and tranquility of knowing what to do next. We also enabled a shareable batch and this screen is also screen-shottable, so it can be used for events. We actually were talking with some people in Europe about using this app for enabling in-person events, but we're still in the process of doing the talk. So a lot of the attentions with COVID has shifted to vaccination. We still think getting people diagnosed promptly is extremely important for disease and others that are out there as well. To build this, we leverage a lot of Amazon's and Google's online tools. So the algorithm actually lives in Amazon SageMaker, which is an AWS cloud solution that enables ease of use of machine learning algorithms. It was tuned and adapted to our use case by our data scientists with Python and R, that are the languages which is written. The backend, all that big data is living right now in an NGINX instance, and it's connected to the React frontend through a Golang API. At this time, we're still looking for other people that are interested in joining us and helping further develop the solution. We have been talking with Novartis, which is a big pharma company, as well as the Health Ministry of Costa Rica to study the potential applicability of this type of solution. Let me know if you're interested or if you know anyone that is. Here's my contact information, you'll have it available. Thank you.