 Hi, it's Monica Wahee here, your Epidemiology and Data Science Mentor and Coach. Today we are going to revisit an important point I make in my online applications basics course and to all my customers and the participants of my mentoring program. And that is that the agile development approach is not a good idea for healthcare and public health applications. As you'll learn if you take this course, the reason is that agile typically sacrifices design steps that are critical to the successful function of healthcare applications. So you might be saying, what's the alternative? Do I have to put up with a ponderous traditional application development approach? And my answer is no way. Of course not. You can instead use my rapid application development or RAD approach. Before we go on, I just want to remind you of a few principles of my RAD approach. The first thing to remember is that it is actually based on the traditional application development approach, not the agile approach. But one main modification is that it does not use matrix management. Instead it uses a very small dedicated team of cross-trained individuals. Also, we don't use the project lifecycle as the basis for project management. I know a lot of project managers hate that when I tell them about it. But to be honest, I have had a lot of success without bothering with the principles of the project lifecycle. I still use a timeline as a sort of gantt. I just make it an Excel, not Microsoft project. One of the main things I like to try to do is eliminate SMEs on my team. I do that by just having SMEs who are cross-trained as techies. So if I was developing a medical records application, I would get nurses cross-trained in software engineering or informatics or some other technical discipline to serve on the development team, so they could be useful in more ways than one. And as I emphasize with the memory disorder clinic database example, Monica's rad is like agile in that it relies on short development cycles. Let me start by explaining the scenario to you. I call my colleague Natasha my forever intern because we keep doing projects together. She's cross-training as an application developer and a public health professional, but she has already had several careers, including a long one in music performance and education. We both live in the same state, Massachusetts, which is actually a very innovative state when it comes to both healthcare and technology. The story I'm going to tell you is about how Natasha identified an open government data or OGD implementation of an online dashboard to showcase hospital acquired infections or HAI rates at hospitals in Massachusetts. She showed it to me and we both really couldn't use it, even though we knew more than the average bearer about HAIs in Massachusetts. So at Natasha's prompting, we decided to do an experiment and see if we could rebuild it together and come up with a better solution. So with my intern, we decided to do something that was one step up from vaporware. Vaporware is where you design an app, but you don't build it. We didn't want to do that because my intern likes to build stuff. She's a very good front-end developer already, which is great because I'm not. So together with me and her as a two-person development team, we sought to rebuild the existing dashboard that we didn't like into a new alpha prototype as you see circled on the slide. We succeeded and that's what I'm going to show you in this video. My intern is busy now. She's off getting advanced degrees, but when she's ready to do the next step of this project, we will do some UI UX studies and revise the alpha prototype into beta and then launch it in the field for testing. Okay, let's start with the dashboard to which we as users objected. It's here at the Massachusetts HAI report site. Okay, see here this 2021 interactive HAI map. Let me show you this. Okay, this is what you get when you go to that interactive map link. All right, so let's say a consumer like me actually knows what an HAI is, which I do. I know that hospitals with high levels of HAIs are dangerous because I could get an HAI myself there and die. So as a consumer, I'd want to choose a hospital with a relatively low incidence of HAIs. Let's see if I can figure that out from this dashboard. Let's start by clicking on this hospital tab. Hmm, it says, select a feature on the map to view details. What do we have on the map here? I guess we have hospitals. Let's pick one over here near the border of New York. Oh, look, it's Berkshire Medical Center. Hmm, but where is the comparison to other hospitals? And what is its rate of HAIs? Oh, here at the bottom, it says, to view this facility's HAI data summary, click on the following link. Let's do that. Whoa, this opens a PDF. I totally did not expect that. Let's see what we have here. Oh, look, it's a seven page PDF with a table of contents in the upper left. I don't see how this helps me decide if this is a risky hospital or not. Okay, I'll spare you the details. I studied these PDFs and I could not make head nor tail of them. I couldn't tell you if any of these hospitals are dangerous or not. I do not know what the point of that dashboard was or what the point of these PDFs are. So we went back to basics, defining who our audience was, which is who would be served by the dashboard, and writing requirements. First, we decided that since we could serve as SMEs ourselves in many ways, we needed to define what kind of SMEs we were, what kind of users were intended for this app. So we decided to make our solution for an audience defined as people living in our state, Massachusetts, who want to compare hospitals and come up with a primary choice list, a go to list for non-emergency needs in whatever region of the state they are in. Obviously, if you have a car accident, you'll just go to the nearest emergency department. The solution we designed is not for that. Our dashboard is for shopping around for a hospital when you have the opportunity to do so. So next, on the right side of the slide, we set up the scope and requirements. We agreed that we might not be able to tell which hospitals are objectively safe or risky, but we could tell relative to each other which ones were the least risky and which ones were the most risky. We decided that this was the scope to be able to make those comparisons easily, and we set up a few requirements as you can see. Then, after we established these basics, we decided what roles we'd both play on the development team, and we went on to develop our solution. I won't go into the details of our Gant, mainly because you can actually read our book chapter to find out whatever you want. But what I will do, however, is showcase our Alpha prototype solution for you. Okay, here is our solution. Now, let's go find that hospital we looked at with the old version of the dashboard, the one near the New York border. What do you notice about it immediately? Exactly. The dot is colored yellow. Let's look at the lower right. Oh, here we can see the legend. But even if we could not see the legend, just looking at all these dots and even guessing, we'd guess that we'd prefer the green ones and try to avoid the red and black ones. But yellow, maybe that's relatively okay. Let's click on it and see what it says. All right, we got the right place, Berkshire Medical Center. Only this time, I think we can understand this information, right? We can see these metrics and they are labeled with essentially the equation used to calculate them, overall catheter infections rate for all ICUs. This is for the urinary catheters and the metric is 2.03 per 10,000 admissions. So you could go and look at the other bubbles and compare this metric. And up at the top, you can see the absolute number of admissions, what is in the original denominator. So you can get an idea of how busy this hospital is or the size it is compared to other hospitals by what it says up here. We thought this was pretty intuitive, but admittedly, we just kind of threw it together. It still needs refinement, meaning we need to do user testing and get a lot of feedback and make a list of improvements. And also, there are probably new data available that we could add. Oh, and Natasha did her homework, which was to at least go around and show our prototype to people like her professors and other big wigs in healthcare policy she might encounter on her educational journey and get the reaction. She told me the preliminary feedback she got was similar to my immediate response, which was to point out that the NHSN data are crampy, so why waste time visualizing them? I guess that was kind of a nihilistic response, because Natasha pointed out that it's possible to improve the data collection, at least for the hospitals in one state. We could conceivably do that as a future project. But for now, all we have is a quickly thrown together alpha prototype. That's what you get with my style of rad. And that's what you get with Agile II. But the difference with rad is that developing the alpha this way was way cheaper and quicker. And also, it's pretty good for a first pass. This is probably an 80% solution right here. Going from alpha to beta could be conceivable for, say, a master's thesis or doctoral project, or even a small local demonstration project. I admit, I went over this case study very quickly. But if you got something out of this, please take a moment to like this video. And if you use either Agile or rad, or even traditional, for your application development approach, please leave a comment and tell us how it's going. And hey, if you want to see the full video for this case study instead of this abbreviated version, please visit my applications basics course online and watch the video for this case study. I made it free to the public, even if you do not choose to take the course. And remember, if you join our mentoring program, the application basics course will be yours free of charge as part of program participation. If you want more information, look for links in the video description. Thanks for watching. Thank you for watching this video, which is part of the Public Health Today to Science rebrand program. If you are interested in joining the program, please sign up for a 30 minute Zoom interview using the link in the description.