 That's great. That's great. And I know that your talk is about roadmap to open source. That's amazing. I think that that's a lot of people found it very interesting. And so right now, okay, I can see people are coming back now. So maybe we will just get started when you're ready. I'm ready. And I'm ready to share my screen. So, right. Here we go. I'm sure you can see my screen. And yes, the today, the title of my talk is roadmap to an open source artificial pancreas and diabetes is monitoring with blast. And if you see a hashtag in here, it's we are not waiting. And it's something that signifies a lot for the diabetes that the do it yourself diabetes community. We are now waiting because we're making things happen. But before I get into the subject, I like to introduce myself. My name is Diana Rodriguez, and I'm a Python developer advocate at Vonage. I'm also a Google developer expert Aussie ambassador and Microsoft MVP. And all those things actually mean that I'm a person who learns new things every day. And I love interacting with the community so you can find me on Twitter at Katoo firey to, or you can go to my website, super die dot dev. And yeah, let's let's talk about who diabetes. I am a non-Hodgkin's lymphoma survivor. One of the side effects. And also one of the things that happened after this battle with cancer was diabetes. It's also a, well, it's part of many things. So when I first got the diagnosis, I felt exactly like either of the drawings in this picture. It felt like it was too much to handle. It felt like I couldn't do it because suddenly the responsibility of regulating insulin for my entire body was on my shoulders. Too much information to process. And I felt like I broke in sides that I wasn't going to make it. Before carrying on, let's go over what diabetes and the types are very quickly. So there are various types of diabetes. These are the most commonly known type one and type two. So we're type one diabetes type one diabetes also known as juvenile diabetes is an autoimmune disease where your body stops producing insulin. In my case, it is still an also immune disease, but it's sort of type 1.5. And it's called latent autoimmune diabetes in adults. There's also type two where your body produces insulin, but doesn't know what to do with it and it becomes resistant to insulin. And there are more types of diabetes like type three when you get your pancreas removed gestational diabetes and so on. And you might be thinking, well, yeah, why all this fuss? Well, what's the challenge? I mean, diabetes is a condition. Yeah, it's an autoimmune disease in some cases. It's an insulin resistance disease in some other cases and so on. We have many challenges. One of the biggest ones to start with things like access to insulin and medication. In the case of yours truly, I live in the United States. And this is from four years ago, but it's a good picture to compare the fact that we have to spend so much money for essential medication. Access to insulin and medication is so costly that nowadays people die because their ration, their insulin, their ration, their insulin. This was in 2016. If I were to go without insurance, I would probably spend between $1,500 to $2,000 per month just to be alive. And I'll leave you that to think in. Then the other huge burden is, well, we have a chart and a lot of information on what the target blood sugar levels are for people with diabetes, right? We're running an automated process manually with some guidelines. And well, what happens when you run an automated process as a human, you make it an error-prone process. Then there's the fact that we have to digest all this information that, okay, when you eat food, part of it is made of carbs or carbohydrates, which turn into glucose and then insulin regulates the right amount of glucose in the bloodstream. When your body stops producing insulin and when it's not using it appropriately, then you have to step in and regulate those processes manually. What happens? Well, those ratios of how much insulin you need to what amount of food you eat are different. Every person has a different ratio and those ratios vary over time. That's more information and it turns extremely overwhelming. And then there's a fact that insulin needs are different for everyone. I might need more insulin if I'm stressed or if I'm nervous, or in my case I need more insulin at night than in the morning and I don't know why. I have a friend who needs more insulin in the morning and so on. And then it's the fact that it's too much and not enough at the same time. We don't know everything and we're truly never under control. Like I said before, when a process that should be running automatically is run by humans, it becomes an error-prone process. And I know this is a long introduction, but I want to stress the ideas, the basic concepts and the concerns and the problems that technology can solve. People, if I don't give myself enough insulin, I can die because I've got sugars. It will kill me slowly, but it will kill me. If I give myself too much insulin, I can die much faster. It's extremely dangerous. Imagine the burden of knowing that one mistake can actually kill you. And that was something that was, it had an emotional toll, a huge emotional toll on me thinking, oh my gosh, if I mess up, I can die. It took me a while and a lot of support from the community to get used to it. And I still make mistakes. There are three main reasons and why I took it to open source to using technology for good. The second and the third reasons are second, community. There are many more who could benefit from any work I can do to help with this initiative. The third, well, I'm a software developer. I know I can help others. I know I can contribute to projects that are already happening and I could inspire others to do so. But the first one, it's not even me, myself. It's her. I do it for her. This is my seven year old niece. She has type one diabetes since she was a baby. And in this picture, she's happy and super proud showing her sensor and her insulin pump. With this amazing technology, she can lead a mostly normal life like every other kid. She can do things that weren't thinkable 20 years ago or even 40 years ago. So although we have all this amazing technology that has evolved, it's still not enough. And I'm sure you might be thinking, well, how can we help? In your context, please. I'll show you. Basically, there are plenty of pumps and sensors like the one I have here in the market. Sensors solve the problem of having to prick my finger multiple times a day to just wearing a device that is transmitting my blood sugar values without me having to prick my finger many times a day. Obviously, technology is not perfect. I have to calibrate. Sometimes I have to check those values, but much less than before. And also pumps solve the problem of having to give myself multiple daily injections. But you remember how I said we're never falling in control. The issue is that pumps and sensors don't speak the same language and they don't communicate between themselves. What's difficult is that we need that monitoring versus that incident delivery to work together so we can recreate as much as we can. And with the technology available, the same behavior that an actual human pancreas would do. And this is where we can step up. We can use tech for good causes. Okay, two things were already resolved. And, and I want to highlight these two initiatives and then show you what my contribution and yours can be. The first problem is monitoring. And that is solved in a really nice way. We have night scout and again the hashtag we are not waiting. We're making things happen. We're making things accessible night scout is a web based continuous glucose monitor. It basically shows glucose data in real time, and it can also predict block glucose values 30 minutes ahead using another regressive second order model. What night scout does it's basically a web a web dashboard. It's a server that reads from a Mongo database containing data from this sensor or any sensor. And then it has alarms that are generated for high and low block glucose values. And it's free open source software. So if you want to check it out in action, you can go to dynamics that super died dot dev to see my blood sugar levels in real time. I don't know if they might be a bit high or a bit low when I'm overexcited or nervous. Actually, here they are. They're quite a right. But it looks like something it looks like this actually. That's a screenshot from a while ago. But we have that great. We found a way to gather the sensor data and to input things like the amount of carbs we consume to record things like if we're stressed or or things like sensor changes, etc, etc. And now we have the ability to run reports with these dashboards and make those accessible to those who are monitoring. Imagine if you're a parent with children who are diabetic, then, you know, you can keep things in check. But what happens? The second step is automating that process of gathering that data and communicating with an insulin pump to make sure that we're delivering the right amount of insulin, or that we're at least as close as possible to an actual pancreas. And that's the second problem that's already being solved. It's automated. And for that we have open APS, which stands for open artificial pancreas system. And it's a safe and powerful tool. It's easily understandable. It basically adjusts the insulin pumps delivery to keep those blood glucose levels within range at all times. So it communicates with the insulin pump to obtain all the details of all insulin doses. And it communicates with that continuous glucose monitor to obtain those values and estimates and it issues commands to the pump to adjust the dosing as needed. Exactly like a human pancreas would do as much as as as possible. I got this quote from the open APS website and as of July 13, 2020, there are more than a huge number of individuals around the world with various types of do it yourself close loop implementations that we know of. And this numbers continues to grow as does the number of options for these types of close loops and and I'm going to explain what looping means. Looping is basically using these tools to automate the process of getting blog glucose readings from a sensor and issuing commands from a pump for a pump to do so automatically like a pancreas would do. With technology like this, that is open source and that is accessible for everyone. We experience fewer highs and less severe lows and more time in range and that is the ultimate goal to have a decent range and to not suffer the consequences of giving ourselves too much insulin or too very or very little insulin got nervous and and then lead a normal life. So, yeah, I have my sensor. I can see that my phone. I also have this web dashboard and it communicates to my insulin pump and I can make make better bolus in decisions. I have the choice to just see the suggestions and not having those commands be issued to my pump, which is called open looping and then to fully automate the process. And I have to tell you since I started with this. All recommendations that were done by by the combination of my blog glucose readings and my pump has been spot have been spot on. And I haven't had any traumatic episodes where I have seizures or where I pass out or I lose vision because I'm too low or because I'm too high or because I'm not in range. But anyway, what's my and your contribution. Well, you see flask in action. I needed something quick. I needed something I could share and have others develop on top. If you want out features, if anyone wants to contribute. I decided to use flask because to me was a right choice choice because of its training features, the freedom minimalism and learning curve. So you'll see two links. There's a repository and there's an application that I will explain. I'm using flask with ginger for templates and material ICSS for a little bit of color and bonus communication API is what problem are we solving on top. Of all these initiatives. Well, the problem is simple. We have all these devices we have the dashboard and the alarms, but you know what in moments of crisis when I sleep. I never hear them. But strangely, I can always hear my phone, which is why the idea was to make my life and the life of those fellows fellow diabetics simple or simpler. People have the same struggles. Maybe you don't hear alarms, but if the phone rings spot on right. So taking advantage of the simplicity of flask and Vonage API is to do this was a good choice for me. But how. So we have range of values. And we have phones. And we have monitoring. So I'm using the SMS and voice API to notify myself and my emergency contacts of irregular situations with those levels. So I grab readings for the last 60 minutes or any given time frequency through a Python thread with a scheduler. If ranges are below or higher than what is configured through environment variables, then it fires and alerts. There's also an alert if it doesn't get data for a set period of time. And in this case, it sends a message to all my emergency contacts saying that my dashboard is offline. The actual solution. In 3d is a simple flask application that allows people to log in and configure their night scout decels and endpoint to grab those readings and the emergency numbers to do so. It needed some protection. So I used Google authentication. And again, why flask? Because it was the friendliest choice for me. I love flask. I have to admit that. In here, I can configure the entries endpoint to grab those readings. My phone number to receive a call if my levels are out of work. An emergency contact in this case my mom, who will receive notifications should I not pick up the phone. And up to five additional emergency contacts who will be notified should either of them not pick up one by one. But let's have a quick run through on the technical decels that may developing this app enjoyable. Easy or friendly actually much better for reviews. We use ginger as the template engine. You know, it's basically HTML and CSS. And we have a parent template defining content blocks to be used by children templates. And these blocks will display the code from children templates. Something that I like to also highlight is things like the URL for function that points to JavaScript and CSS resources in a static folder. Which is a very comfortable and elegant way to point to CSS libraries or any scripts you need to execute for your for your views your front end right. I'm going to show an example of this with the login screen you saw before previously the screenshot. And if we see, I like how the content blocks are used we were extending layout or HTML. So this is inheriting from layout and then the renderer will load what will load layout with whatever variations we add to this file. And these variants will be defined within the blocks we see layout for example I'm using the head block in this case. And well, I kind of filled it up with a little bit of coats for you to see in context. If you go on the Python side of things. One of the things I, I truly love about Flask is the fact that it is easy or friendly to read it's friendly to understand and it doesn't matter. If you're a beginner or how advanced you are in development. Something to highlight the basic is the beautiful syntax for root decorators where you can specify the roots and the method if applicable and then write your logic. So yeah, I wanted to make this as friendly as a glass of water friendlier than free stuff. But this is not even the highlight of it. If you really want to take a look on the nitty-bitty grits of the code. I already wrote full step by step tutorial on how to build this application. You can go to next month.dev forward slash night's go and you have all the details. The most important part of this talk. After giving you all this context on diabetes and showing you a little bit of what has been done and what I personally did to improve this community and improve the lives of people like me, like my knees. And all the people who are affected by diabetes worldwide is our challenge. We have a tech for good challenge. How you operate head to next month.dev forward slash Europe Python 2020. We are basically rewarding your contributions for this application. And it doesn't matter your level of expertise or how many years you've been developing. It's open for all those who want to contribute. And in this challenge, we're asking your help to find features or ways to improve this application to improve the lives of those who are benefiting from it. People like me. So we have prices and we have a number of indications and repository with a really good documentation and all the endpoints are documented. Everything is laid out beautifully and the frame is way possible for you to take part in. So there we go. Also, tomorrow we will have our booth or our bondage booth. Come and see us. We can talk about all these APIs. We can talk about the challenge. And if there are any more questions, I can answer for you. We'll be there. I'll be there and I'll be happy to talk to all of you. If you would like some more resources and learn more about these initiatives, I suggest you to follow a couple of Twitter accounts like Sarah withy. She was one of the people that supported me through this journey on coming into the DIY diabetes community. There's Scott Hanselman. He's very famous worldwide for being a pioneer and automating these processes and he has amazing information as well. There's also the night scout websites and the open API open APS website. And here you will find more information on these initiatives and how people are using them nowadays. And then of course, if you want to know a little bit more about me and about what I'm doing. You can go to Superdye.dev and you need a copy of these slides. Here's the URL. Again, Superdye.dev is my website. And if you want to chat on Twitter, I'm available for 2582. Come and see us tomorrow in our booth. And I'm excited for what this challenge will bring. I'm really excited to see how can we as a community improve this application that I'm currently using and has saved me from quite a few tricky situations. And this is what I have for you today. Thank you so much for this opportunity. And I hope that you feel inspired and motivated to make things happen and to use your knowledge for good causes. Thank you very much. Thank you so much. Diana is really good. Like it's using open source using Python to take care of yourself. This is amazing. And so we have just like a few questions popping in so maybe we have time for one or two questions. So, so one of the viewers asked like how did you configure cyber security for the app. Well, it's work in progress. I mean if anyone accessed the app, it's not like they can issue any commands to like no one could kill us. It's just information. It's basically entries of glucose levels. And that is something that I would like to see as a contribution on how can we implement better security. We're basically just using standard, you know standards as much as possible. So, so yeah, so if you'd like to contribute with more authentication features or if you have anything else to contribute in terms of cyber security I would love to see that personally. Right, so there's another question. So I think this is the last question now so like how costly is the hardware. Well, the hardware, the hardware is not costly or it will it is in a way, we have the privilege of being most of us have the privilege of being insured. I would say the sensors are around $300 for a 90 day supply. We would basically have worked in a DIY community to find ways to connect to the sensor APIs or through their dashboards. The Dexcom has a private API and we can access that we can bridge it to like that we can share those details. There are other sensors like the freestyle liberate, which are less costly, and in most countries in Europe are subsidized. And there's an IOT device that goes on top because the freestyle liberate is an NFC sensor it doesn't send signal on, you know, via Bluetooth. So there's a small IOT device that you put on top that is tapping constantly this device and then we can connect it to any of the open source dashboards. The cost varies in, in different countries, I think the United States being the most expensive of them all, I'd say $300 to $400 per 90 days out of pocket for Dexcom and around 120 from well that was a thunder from for Libre in the United States. Right. So yeah, thank you so much. I think if you have more questions you can go to the, you know, the channel for the talk to keep on, you know, the discussion maybe ask more questions. You know, ask Diana or continue the discussion or how to contribute we can ask her. So thank you so much.