 Okay. One, two, three, four. That sounds really good. Oh, I have a deep resonant voice. Well-modulated. Okay. I couldn't figure it out. I love to turn that on. Okay. And I guess there's a little bit of a danger of feedback if both of these are on, right? Oh, you're going to be here moderately, so you can use that. Uh-huh. Uh-huh. I'm sorry. Okay. Yeah. Yeah. No problem. I did that yesterday. You know, to that thanks to everybody for coming, let's say again. I'm going to hang out here. La-la-la-la-la. Okay. One, two, three, four. Hi. Welcome. Hi. Welcome. Okay. It's 1131. I'm going to get started. Hang on just a second. Let me see if I have to. Okay. Well, I don't know how the recording works, but I'm just going to begin. So this talk has quite a few links that you may wish to follow. You may wish to use your camera on that QR code to get that. I can take time if somebody wants to do it. Hi. Welcome. Thank you. So my name is Robert Reed. This talk is called Free Spirit Code, Free Respiration Ecosystem and Why Scale 19x Should Care. I'm the founder of Public Invention, which is a US 501C3 Public Charity. You can email me or follow me at these things. We have a small audience today. I'm very happy to be interrupted. This talk combines software, hardware, medicine, and policy. And so there's a real danger that I may miss. I'm not familiar with this audience. So there's a real danger that something could go wrong. So I would like to be interrupted if I've lost you in some way. I'm happy to adjust my talk to what people want to hear about in this case. So I'm happy to take questions. Public Invention is the idea of making open source hardware inventions, which I consider to be an extension of the open source software movement. Our motto is to invent in the public for the public. We have a website. And I'd like to talk a little bit about the timeline of Free Libre open source movement. So Ms. Black's talk was really hard to follow. I thought that was a great talk. That was really a fantastic talk. What I'm talking about here in some ways fits into what she was talking about. She was talking about the history of open source software. I'm attempting to talk about the extension of that idea to open source hardware in certain ways. So as she mentioned, the Free Software Foundation and also the GNU Manifesto were written in 1985. Linux was produced in 1991. I first attempted to install it in 1992 over Modem with the Iggdrasil release. It wasn't super successful. But a lot has happened since then. About 10 years ago, people started talking about open hardware. And when they said that, they mostly meant integrated circuits and chips. The RISC-5 Foundation was founded in 2015. To me, it seems like in just the last year, they've had real success and really have become mainstream. I don't know everything about it. And then I have been along with a number of allied nonprofits working on open source medical devices inspired by the pandemic. Now, of course, it makes sense that this is the order in which human society would deal with these problems, right? We've had notions of fair use and open source text and the expiration of copyright on text for 100 years or more. And of course, that's because it's relatively simple. In the case of software, we have the GPL and the epheral license and the permissive licenses such as MIT. And now, even though this remains complicated, it's sort of well understood. Hardware is much less so. There are attempts to produce licenses for open source hardware. In particular, public convention uses the CERN, a strong reciprocal open source license, or recommends that. But a competent intellectual property attorney will tell you that there are grave limitations on how much that copyright can really do for hardware devices. It really covers the design documents rather than the hardware devices themselves for which we have patent monopolies, something called patent protection. Public invention never seeks patents. I'm antithetical to patents. But patent law gives both ways of forming monopolies on hardware devices and also limits the monopolies which can be formed on hardware devices. Medical devices are even worse because they are hardware devices which in some cases are life critical and are regulated in the United States by the US FDA. Now, I attempt to do a lot of international work. Other nations do not have to respect the US FDA, but I'm going to pretend that they do because in practice they tend to follow what the FDA does and it's a stand-in for the regulatory bodies of other nations, even though, for example, low and middle income countries in Latin America, Asia, and Africa do not have to follow US regulation, but in practice they often sort of use it as a template for what they're attempting to do. Okay, so I'm just going to give you a little bit of history about what I personally did. In March of 2020, when the pandemic hit, you guys may remember, there was an expected shortfall of mechanical ventilators and this was expected to cause tremendous death. It turned out that that was not true, but it was not a panic. It was a reasonable assumption based on knowledge that we had on how to treat the disease at the time. In Northern Italy, we learned some things about the disease and the doctors very quickly changed the standard of practice, so it turned out we didn't need as many ventilators as we thought. It's also the case that in April and May, we did not know in the United States that social distancing was going to work. It could have been that it wouldn't. Turns out it does. It's politically expensive. It's difficult to socially distance. People prefer to interact with each other, but if you walk a place down, it will shut down the transmission of COVID. So for all of those reasons, the need in the United States for a million ventilators, which was legitimately predicted, did not occur. At that time, however, by my count, about a thousand humanitarian engineers who were mechanical engineers, electrical engineers, and software engineers, in some cases also supported by medical doctors, clinicians, respiratory therapists, started working on open source ventilators, and they also worked on other things. Now this was a noble effort, but almost a complete failure in terms of saving human life. However, other aspects of it were not. For example, almost everyone needs personal protective equipment, masks, gloves, gowns, and so forth. Another organization that I worked with, open source medical supplies, did not produce but sort of managed and oversaw the production of 71 million pieces of personal protective equipment. Now my belief is that probably slowed down the infection and probably saved some lives, although you would never be able to prove it, okay? A much, much smaller number of people need therapeutic oxygen, which is a problem worldwide right now, especially in low and middle income countries, and noninvasive ventilation. Only one percent of the people who get infected with COVID-19 may need mechanical, invasive mechanical ventilation. I'm going to talk about that because public invention is building such a ventilator for different reasons. So public invention has created the open medical technology manifesto, and if you go to the change.org petition, and I'd be happy for you to sign it there, you will see that it has a 10-point program, okay? But in a nutshell, my argument is open, shareable energy here at scale. There are probably 250 people here, maybe 500. There's a lot of energy, and this is only one of many open source and Linux conferences in the world. There's a tremendous amount of sort of intellectual energy, and dare I say love, devoted to open source software. What I'm attempting to do is to build a movement so that we have that much energy applied to open source medical devices 10 years from now, okay? And it's going to be kind of a long road for reasons that I'm talking about here. Now, thank you everybody for being here. Some of you came in late. This talk has a lot of links. There's a QR code. If somebody wants those links, I can go back to it briefly. If you want to do that, does anyone want those? Okay, you can look at it. Some people might want to look at it during the talk, but that's okay. Okay. So I'd like to talk specifically about the thing that public invention is doing. So public invention is a very small, low-budget U.S. charity. It's mostly me with the help of some volunteers, and we are interested in creating a free respiration ecosystem. So there's nothing more boring than hearing someone read this, but I'd like to just read the manifesto for this. The COVID-19 pandemic has demonstrated a clear and present need for complete free Libre open source, easily repairable, widely usable, safe and effective respiratory support medical device ecosystem. And the ecosystem is a part of it that I think almost every open source advocate and open source programmer will understand, because in the time since I learned programming 35 years ago, programming has sort of changed from mostly writing code to mostly figuring out how to use code that other people wrote, and occasionally writing a little bit of code to tie those things together. Okay. We understand the word ecosystem. We don't even use it really in open source because we now assume it, but doctors and medical technology doesn't really work this way. So the problem we're facing is that fragile international supply chains kill people. Okay. In the United States, this is usually not the case. We usually have the medical equipment that we need. This is not the pressing problem in the United States. In low and middle income countries, it's a very different story. Okay. Public Convention has put on conferences. The keynote speaker at our last virtual conference showed a picture of a room about this side, and the entire part of it there was stacked up with broken oxygen generators from Africa. Okay. So often in Africa, for example, people donate equipment, but the ability to repair the equipment is not present, and those devices are almost always proprietary because right now there are almost no open source medical devices. And furthermore, they're not made to be repaired. And they're not placed in a place where you have a culture that is practiced at repairing them as well. So for all of those reasons, it's a major problem, and real people die because of the lack of equipment there. So the solution is to create modular open design respiration devices. And to me, the modularity is really important because it allows small businesses to comfortably start to become part of this ecosystem as opposed to the way things are done right now. Do we have any questions about this before I continue? Okay. Please feel free to interrupt me because I'm bringing in a lot of different concepts here, and I know it may not make too much sense. But modern hardware devices are almost always software devices as well, in the sense that sophisticated medical devices normally have some kind of a microcontroller and normally run software. So an embedded system is now an interplay between software and hardware, okay? So those of us who believe in open source software can easily believe that the same advantages will accrue to open source hardware. But of course, there are differences. First of all, hardware does not have zero marginal reproduction cost. It's never going to be as cheap to make any kind of as it is to make software. Now in the last 10 years, the cost has decreased tremendously with, for example, Arduinos, 3D printing, CNC machines. The cost of manufacturing any kind of device has actually gotten much lower, but it's never going to be free. And it's never going to be free to make 100 of something. Ms. Black was talking about the need to have many eyes on the problem, and she sort of disparaged that as something that you could not 100% trust. Of course, she's correct in some ways, but on the other hand, I think most people believe open systems in the end are more trustworthy. That is not true of medical devices, and for example, there are people whose lives depend upon the software in pacemakers embedded in their body, and they cannot view that software. If there is a bug in that software, they may die and they're out of luck. They have no access to that software. If, for example, the company that makes it ceases to maintain it because it goes out of business, they may be completely out of luck. Now, the other technical difference between making hardware and software is that if you go to the expo and thank all the sponsors of scale for doing that, you'll see that automated unit tests are a fundamental way we make and trust open source software today. Everyone can easily and repeatedly run a set of unit tests. Even if those unit tests do not cover all of the things that you would like to test, they give you great confidence because you can read the test and see what they are testing. You will always have to do mechanical tests and mechanical devices, and it will never be as easy to test them that way. You may still, however, have a culture of repeated tests in the response to design changes, which does not exist today, but you will always have to sort of take some manual steps. Somebody's got to plug something into something and run the test. Okay, so in embedded systems, there are hardware and software, and as everybody here understands, you've got to have something like GitHub. You've got to have source control, and you've got to have a so-called forge of software for people to share and collaborate on the software. You have to have code cooperation and review processes as Ava Black talked about. You also have to have an understanding of licensing. Public Invention has its own guidelines for that. We use different licenses for different kinds of matter. For mechanical inventions, we use the CERN license for software. We tend to use a FARO for documentation. We use CC0, et cetera. We also need something that was brought up in that talk, but not completely explained, is we need a social reward for contribution and a potential financial reward for contribution. Now, today, there is no very definite financial reward for contribution to an open-source software system. There is occasionally an indirect reward in that it improves your resume, you're more likely to get a job. A small number of people are paid to be professional open-source programmers working on an open-source project, but most people are not. However, there is, in some cases, a social reward in the esteem of your colleagues for making such contributions. I personally believe that needs to be expanded in the next 10 years, and we need to provide more financial rewards for open-source developers of software, but also of hardware because hardware costs more. Now, this conference is about Linux. I'm a huge fan of Linux. In the ecosystem that we're going to talk about, we use Linux for our data cloud, for our so-called IoT, which is the worst name ever, the Internet of Things, where we monitor things and we publish data via UDP to a public data lake. Personally, we don't do that on the actual devices we're building. We tend to use Arduino-level work. Even though you can run Linux and use a Raspberry Pi-type device, it's generally not worth the complexity of dealing with version management and upgrades and security holes and all that kind of stuff. And in general, the computational needs of the embedded device are so simple that you can use microcontrollers and microcontroller-level software. If anyone wants to talk about that after the talk, we can go into it. So open-source medical devices will never be free of charge. Today, people like Richard Stallman and myself will insist that free means free isn't speech, not free isn't money. But it's also the case that open-source software tends to be cheap. And often it is free, and that's one of the great advantages of it. That will never be the case of medical devices because the marginal cost will never be zero. It is also the case that open-source medical devices will not be free of regulation. In the United States, you can build whatever you want. But if you sell it for a medical purpose, without US FDA approval, you are violating US law. OK? So we have made a social compromise when it comes to cosmetics, food, drugs, and medical devices that those devices should be regulated because there are so many horrible stories about lead in cosmetics or people embedding goat testicles into their own testicles and creating all kinds of problems that we believe it's worth it not to have quacks so that we have FDA control. That is not the case in software. There is no regulatory body that says you can run this software. You can't run that software. But there probably always will be in medical situations, and I consider that to be a good thing. Nonetheless, if we can make open-source medical devices, they will be cheaper and more available than they are today, where there are a lot of financial pressures which make them highly monopolized. OK, so I'd like to talk about a technical example. Along with an anesthesiologist from Australia, Dr. Eric Schultz, public invention has developed this universal ventilation model. Is there anyone here who doesn't know what a mechanical ventilator is for a human being? OK, so I won't go into it too much. I hope none of you ever are on a ventilator, or one of your loved ones is not on a ventilator. It's very unpleasant. But all ventilators from a mechanical point of view are simple in their function and can be placed on this diagram. They all produce air and or oxygen, a controlled mixture of air and oxygen. They all sense that flow and pressure in order to very carefully control the amount that goes into your lungs. Too much goes into your lungs. Your lungs can be damaged. If not enough goes in, you die of asphyxiation. So sensing and controlling this is the fundamental job of a mechanical ventilator. The pressures used in breathing are tiny. Almost always less than one PSI. I put 100 PSI of air in the tires of my bicycle, right? It's a very low power operation. But it has to be controlled very precisely. For that, every such system has to have a controller. And every controller has to have a user interface. Why? Because doctors have to control things based on the condition of the patient. Sometimes that condition is simply one patient is larger than another patient. But sometimes the disease is progressing. And over time, a change has occurred within a single patient. So often, it's simply a knob. The more modern and more expensive ventilators use a touchscreen with very sophisticated controls that let you set up all kinds of stuff. It's also the case that doctors want to view a lot of data about the condition of the patient, which can be sensed at the level of the ventilator, OK? Now, today, no one makes a modular ventilator. No one makes anything even resembling a modular ventilator. In general, ventilators cost between $20,000 and $50,000. And there is essentially no sense in which they're open source or even open in any of the definitions of openness. Nonetheless, we can contemplate a long-term program of attempting to build a modular ventilation system. OK, now, I'm sorry that due to some circumstances on my control, I didn't bring my machine. I think Mark and Daryl are going to, in the next talk, show some of their equipment. But I have equipment, too. And believe me or not, it works. And many of these components have what I would call the beginning of an open source solution. Now, none of these are FDA approved, and we're not considering FDA application for any of these devices. The little green box here, the Ventmon, is sort of the most successful product which public convention has made. And I say product because we don't sell it. We got $2,000 grants, one from the Mozilla Open Source Foundation and one from Protocol Labs to make and give away this device free of charge to engineering teams at the beginning of the pandemic, which we did. And that device has a tiny little screen on there. But the screen is really just to show you that it's working. Really, what you do is you use your serial port, you enter a Wi-Fi password, and it communicates through UDP to a public data lake. And that produces the display, which you can't see very well here on the side. This is an interactive, real-time, pressure and flow display that computes all of the clinical breath things, almost all, which an expensive $50,000 Drager ventilator will produce for a patient. Yes, ma'am. Good question. It's in the browser. It's a web app. And so you just point your browser at it, and it shows it with about a one-second delay from the reality as the UDP packets go to the public data lake and then a retrieve via an Ajax call into your browser. And so this immediately, a developer will think, OK, well, this is rather interesting because those are now completely separated. And they're completely open source. So for example, if someone doesn't like my software here, and I actually write this, they could write their own. And it could literally use the same data lake, right? In the same way that open source sort of opens up modularity and competition for these possibilities, modularizing the engineering parts of this become possible. Now, this is all setting aside the question of whether it's really safe to do that with a patient. There's a lot of issues that have to be involved there. But certainly from an engineering point of view, it's relatively easy to understand. The box on the left here, which is quite large, we're making a smaller one, actually is an air drive. It is powered by compressed air and compressed oxygen. It mixes them in a chamber inside there. And the three little ports here are what you need in a breathing circuit, which is similar to what Mark has over here. It uses a standard 22 millimeter adult airway. And it goes to the patient. And it has a patient inflating valve. Now, so this is the mechanical part that can fail. Now, if a ventilator breaks and it doesn't break in software, this is the part that's going to break. This is the part that's going to need to be replaced. By making it modular, you open up the opportunity that another firm, possibly a local firm in a low and middle income country, can make that air drive but use all of the rest of the modules. For example, suppose, let's say, Egypt, they know how to use motors, make pistons, they can make an air drive, but they don't have the software sophistication to use the rest of the thing. They can just reuse all of the rest of the components and make their own air drive for the system. So I believe, for example, the world needs a reusable, free-libre, digitally controlled air drive. Today, there are fans. There are even digitally controlled fans for which there's published a nice flow against back pressure curve. But no one has made a standard, like a plug-compatible standard where you say, this will produce a medical quality of air according to an API which I send it, which could be used in a mechanical ventilator. If we had that, we would open up the possibility of multiple firms making air drives. Now, why would that be important? It's important for competition and reduction of price, but more importantly, it's important for supply chain resilience. Suppose, as happens, today, we're dependent on a Chinese firm to make something that goes into that air drive. If China chooses not to export it, then that could bring to a complete halt the production of the German-made Drager mechanical ventilators. It would be just the end of that supply chain until it could be redesigned. If we applied the same, we would never stand for that in terms of our open source software. We said, well, of course, we're not going to do that. We'll just have a thing. And if, for some reason, the person who's maintaining that chooses not to maintain it, well, someone else can just write it or do a different thing. Now, this is a little difficult from the FDA's point of view, in the sense that the FDA doesn't let you just open up a medical device and change it and consider it still to be an approved device. So it requires policy making and legal discussion in order to make this work. But the FDA already has notions of accessories which sort of make this possible. So there are ways that that can be handled. In fact, with a woman named Sabrina Merlot, who works for open source medical supplies, I have proposed a policy change, which the Biden administration opened the door to allow the FDA to clear designs. So it's important to understand that right now, the FDA does not approve designs in any sense. They approve firms to make designs, which allows them to market a particular design. There is no sense in, there's no legal structure in which the FDA says, this is a good design no matter who makes it. And that's not a good design no matter who makes it. That could be changed. I personally don't think, and I don't, I haven't heard that anyone has looked at this. It's probably like a lot of things that's probably gonna take 10 years for that to happen. Now it's important to understand that large firms will incorporate and improve free designs. So for example, if we had an air drive, there's nothing keeping Drager from using it, right? Like Drager can make money from the free design of the air drive. And that's very important. This is not an anti-business approach. I expect it to lower monopolization and lower the prices of medical devices in general. But there's nothing keeping large firms from making this and contribute to it as long as they don't violate the licenses. Okay, so then large firms should put their energy into making features that are expensive to test and design, which small engineering teams can't do. But in terms of just producing air of a particular flow at a particular pressure, there's nothing particularly tricky about that. So large firms should be rewarded for making what I would call luxury ventilators, which offer features which are not available in plain vanilla ventilators. But of course that makes perfect sense. That's what they get, that's their business model. They make a lot of money from that. If all you want is a plain vanilla ventilator, that technology has been understood for 20 years. There should no longer be a monopoly on that technology because it's relatively straightforward. Okay, so how do we make this real? And I know there's a lot of things here. I have a slide on each one of these, but I kind of want to tell you twice because I think it's sort of important. And now we're in the realm of my personal opinions about these things. This is more a policy than anything else. We need a library of proof of concept designs. And I say proof of concept because that's very different than manufacturing ready designs. And we need to start making a distinction about that. Funding for creating and documenting manufacturing plants to the point of being startup ready. The third party testing organization. And that's something that if someone wants to talk to me after I talk, I'd love to talk to you. This is kind of my pet project. A culture of publishing test data which absolutely does not exist in medical devices today. A license that forces publication of some regulatory application documents at the time of regulatory request. And along with Mark Jones and intellectual property attorney that is on the board of public invention, we have created a draft of such a doc license. We call it sunlight regulatory. I know people say there are too many open source licenses already but I'll explain why we feel that's necessary. Very, very important is to have standards. And I think everyone here will understand standards are the way that multiple teams in multiple nations are able to cooperate and make modular systems. And in emphasis on modularity and repairability for supply chain robustness. So there are already proof of concept designs. Now in the ventilator space, public invention made a spreadsheet that eventually had over a hundred entries in it of all the open source teams attempting to make ventilators. Many of them were frankly doomed from the start. That they didn't have the knowledge they needed to make an effective ventilator. I would say maybe the top 10 or 15 on that spreadsheet more or less made an effective ventilator. I don't know if any lives have been saved on the basis of any of those. So I consider a very important thing that humanity did in the pandemic. A bunch of people came together, tried to solve this technical problem. It turns out the technical problem kind of passed us by before we solved the problem but we learned a lot from doing it. And if we had to do it again, we could do a better job next time. So proof of concept is not the same as startup ready. And why does this matter? University professors have a strong incentive to produce a proof of concept design. They can write papers on that, they can get tenure, they can improve their prestige. It takes a lot of work to go from there to something that's manufacturing ready where you could really create a startup firm. Okay, who is gonna support that work? Well, if you're trying to create a startup, normally what you do is you get a couple million dollars and you do it yourself and then you keep it secret and that's the way business gets done. An alternative is to have NGOs like the Gates Foundation, for example, financially support not just the development of a proof of concept, but an actual manufacturing ready design which could in theory be manufactured in many different nations. Okay, in the case of something like a ventilator where the technology is relatively well known, this is a relatively modest price for the benefit to humanity in my opinion. A third party testing organization is super, super critical. So you don't see it that much anymore, but it's still the case that electromechanical devices on the back, they often have underwriters laboratory approval of it. That's an example of kind of a consumer reports of particular devices. Underwriters laboratory focuses on electrical specifications. One of the problems with open source stuff is you can't trust engineers in the following way. I consider myself a very trustworthy person, but I'm emotionally committed to my work. Don't call my baby ugly, right? There is a danger that I will kill your loved one by saying my ventilator is ready to be used. I want to save lives. I want to be a hero when it's not ready. Okay, and it's not because I'm trying to rip you off, right? It's because I want to be a hero. So we need a third party that's not related to me to analyze the ventilator and perform all the tests on it and to give it a stamp of approval and say, indeed, this is a high quality instrument. Right now, there isn't any such organization. Now, I believe this could be a distributed laboratory, right? It doesn't need to actually have a building, right? It could be distributed. We could say, well, Daryl's made a paper, so Rob contested it because Rob doesn't like Daryl, and Daryl contested his ventilator because Daryl doesn't like Rob, and we trust that that happened. We don't have to have like a dedicated building where we have test machinery for it, but we have to have a repeatable test procedures and those procedures have to be documented. So there are some cultural things that we need to work out. A culture of publishing test data is really important, and this is something that the software community just does. You have automated unit tests and anyone can run them. This is not true of open hardware. There's no standard for how that those tests should be performed and how they should be reported and how they should be documented and how they should be organized. So I think we should work on that. The sunlight regulatory license is perhaps worth mentioning and we need help with this. If there are any attorneys who want to volunteer to work on this, that would be great. The basic idea is that right now, there's only one company that I know of called Tidepool. I don't even know what device they make that has published their entire FDA application. If you create a startup and you wish to get FDA approval, you have no examples or templates to work from. You can read the code of the FDA, which I have. In addition to doing that, you normally pay about $200,000 to a consultant to assist you in going through the FDA process. Now I personally, there's a game called Dungeons and Dragons, which represents ethics in a two-dimensional way. It has good and evil and it has law and chaos. And I personally believe in the medical space you have to be lawful good. And the motto of the lawful good person is kicking ass and filling out all the proper forms. In this space, you have to do the paperwork. Now, when I was much younger, I didn't want to do the paperwork. Like many computer programmers, I'm like, I have paperwork to Bosch, I'm never gonna do that. You just have to fill out all the forms. So we need to develop some kind of a culture that rewards people for doing the careful work in doing all the paperwork that the FDA demands. This license is an attempt to create a creative commons of FDA applications so that people can see them and use them. So the basic idea is if you take a public convention design and you use it so that you're building a device based on something that we have developed, at the time that you apply for regulatory approval, you are legally required to publish your application materials. I have no idea if this will work. Once again, it'll probably take 10 years. Okay, so there's an open secret here that's pretty important, and that's to make open standards. And this may be so obvious, but I just want to repeat it. Even as Ava Black said, what we're doing open source software exists because we have HTTP, HTML, JSON, APIs, OAuth, SSL, IEEE, floating point standards are very important for hardware operation. In the realm of hardware, there are protocols like I2C and SPI, which are electrical level protocols. You know, and when you're dealing with electromechanical devices, you have to include those things. We need respiration standards. Now, public convention has created two respiration standards, one of which is fairly well developed. The public convention respiration data standard. And it's what allows the transmission of data from the Vintmon to the software that draws the little squiggles representing pressure and flow. And in fact, although it's not glamorous, it's the thing I'm most proud of. Like, all of the things that I've done in the last two years, defining and versioning that standard is probably my greatest contribution, even though no one has noticed it yet and no one has used it. I believe that that's actually true. So I'm gonna go a little faster here. So this diagram represents why I call this the free respiration ecosystem, okay? Oxygen concentrators, ventilators, BPAP machines, CPAP machines, Pappers and back valve mask monitors all do the same thing technically. This is why it's reasonable to call this an ecosystem. They all move medical gases in a controlled way, okay? So they all kind of, even though they're all different, have an underlying technology. You have to produce, move and measure and control medical gases. Therefore, there's an infrastructure which consists of standards, most importantly, but also software, which can be reutilized to make these things. By doing that, it enables a production of hardware like an air drive or a dryer or a humidifier that could be plugged together to make sophisticated medical devices in a supply chain resilient way, okay? And then what you eventually have, you have free Libre teams like Public Convention, helpful engineering, open source medical supplies, Cosmic and Canada, various others, using variants of free Libre open source software in embedded hardware and monitoring things and eventually providing them to sick people, okay? But there's a missing piece here. You have to have FDA approval. And in practice, these things are gonna be delivered by firms. Now, at the beginning of the pandemic, people maybe thought, no, it wasn't true. My mom's gonna die of COVID. I'll make a ventilator in my garage for my mom. It wasn't ridiculous. It turned out not to be correct, but it wasn't a silly thing to do. But in practice, firms are gonna have to do that. But we can make a business-friendly approach to utilizing all of these standards. So how realistic is this? Not very, but Public Convention with less than $100,000 in partnership with some other nonprofits like helpful engineering has already made the Vintmon tester device that analyzes flow. VintOS is a, it really is not an operating system, but it's an Arduino platform for any microcontroller that does a lot of what you have to do in a mechanical ventilation system because it's almost all the same, no matter what the hardware is. You just build hardware drivers in. A software person would sort of understand that. PolyVint is our attempt to make a ventilator. We are no longer attempting to make a medical ventilator. We're making an educational ventilation platform. We hope it will be used in a classroom at Rice University this semester. So we're still working on that, even though we've retracted our ambition. We've made standards. We have an oxygen concentrator. That project died. As Ava Black said, I have a wonderful engineer in New Zealand. He got a job working for Rocket Labs and his daughter was born. And because of those things, that project has stopped. So not very many people worked on all this. I personally wrote the Vint display software. At the public convention YouTube channel, you can show demos of all this stuff. I wanted to bring this stuff, but for various reasons I didn't bring the hardware, but you can see demos of that working. So in summary, can we do this? I think we can. We can create and establish standards. We can create libraries of hardware designs. We can move to a more mature model where we also create designs for manufacture. Designs that are ready for manufacture, which are a higher level of design sophistication than a proof of concept down at a university. And we can create a free culture of FDA applications, which would make a tremendous difference if we could do that. And thereby, I hope, over a decade, maybe two, create a culture of open source medical devices which truly democratize important therapies and create universal business opportunities. Now, what do we need to make this happen? Public convention personally needs highly skilled volunteers. And I have found that if you're not willing to work six hours a week, it's not worth my time talking to you. Because I've worked with dozens and dozens of volunteers who say they're willing to work six hours a week and then they're not. And I spend hours and hours explaining things to them and it doesn't work very effectively. So unfortunately, these kind of sophisticated embedded systems where it's not just software, like you have to buy some mechanical stuff an Arduino or a transistor or whatever it is, a fan, and you have to work on it, requires a level of dedication even higher than contributing to an open source software project. That's a real problem, right? That's gonna cut down the number of volunteers probably by a factor of 10 or 20. And we need people like Mark Rodin maybe who can lead entire teams, okay? I can usually find good engineers, but I can't find engineers who are willing to lead an entire team. So if you're one of those people, maybe we could talk. Public convention could also use $400,000. If we had that level of money, we could start paying people. I would like to pay an executive director. I do not draw salary from public convention. Almost nobody gets paid. I do have an administrative assistant who works five hours a week. I pay her. But we need a level of financial support which almost no open source projects are getting right now to make this a reality. But public convention is not that important. What's important is a movement, right? Public convention is a US 501C3 public charity, right? But whether it or I stay alive 10 years from now is not super important. What's important though is that the world and we can be the leaders of it, develop a culture of cooperation. In particular, during the pandemic, there was a lot of the not invented here syndrome which was understandable. Everyone was afraid. Everyone was working in their own basement. They didn't know what to do. They didn't know how bad it was gonna be. But there was a tremendous wasted energy, literally 100 teams. I say that because I have a spreadsheet with all of them on it. Tried to do the same thing. Like there's like 100 people trying to, you know, drink the same glass of water. It was a silly way to organize things. It was understandable under the circumstances. But in the future, we need to do a better job. Finally, we need technical leadership. And we also need policy leadership, both at the legal and the FDA level. And we need nonprofits like, for example, famously the Gates Foundation, but other kinds of large foundations, MacArthur Foundation, Ford Foundation, Sloan Foundation, those kind of grantors to understand that they will have a high return on investment if they fund teams making open source medical devices. And I think I'd like to stop there and take questions, if I may. I'd say one thing real fast before I open the floor to questions. I wanna point out something that Robert sort of glossed over because I don't think he realized the power of it. But if you go back to that graph that you had on the bottom left corner that was really tiny and small. One of the things, so Daryl and I and Sean, we all started to also work on ventilator stuff. We also got out because we said that it was gonna be too crowded of a space. But one of the things that our doctors told us that they absolutely desperately wanted was the ability to have a remote monitor at the operating station flaters. And they could not get it out of a Puritan, Bennett. They couldn't get it out of their 50,000 ventilators. These guys, that Ventmon is maybe $150 worth of parts. Maybe. Maybe. 200. $200. That graph, what it does is it enables somebody to monitor a patient remotely. So they're no longer having to go into a room, put on the PPE, go into a room, screw around with the patient, check to see they're okay, walk out of the room, get rid of all that PPE, put on a new set of PPE, go into the next patient's room. Cause you have to do that cause you have no idea if your PPE is covered in COVID and then you're gonna go and spread it to the next patient. That would allow people to have a remote monitoring station of all their ventilators and then only don PPE as necessary. So there's already innovation happening in the space that the large medical device manufacturers haven't covered, right? And they just got that as a, just sort of like that Ventmon is a really cool device. We tried to use it in some of our stuff and that remote monitoring capability, phenomenal. Like it really is a game changer if we can get it out there, Darrell. So speaking of large corporations making events, there is a $50,000 event right now. That's a pretty standard event that is used in the hospital whose remote monitoring station is basically a screen that's attached by a cable. The company has made it so that it senses if the insertion of the cable is at the right depth on the on the monitors or on the vent side. If it is moved slightly, it shuts down the entire machine. It's a bug, something that you would expect them to fix. Going on two years now it has not been fixed. So they don't use those wires to do the remote monitoring, the monitor in the window because of that. Just to give you an idea, this is really important, the idea of having innovation because if you don't work in the healthcare industry, you probably don't know that there's actually not that much innovation going on oftentimes at the corporate level for hardware and software in general. Updates don't happen very often. And the companies don't have any accountability for this because as long as it's FDA approved, they can sell it. So just want to give you a little background. I also wanted to point out one other thing in terms of funding sources that you mentioned too for startups. One other one that I was thinking about is sort of the government grants like SBIR grants that you could apply for if there were more of those coming from a ivory-towered academic as myself. Those are the things that we also look at in terms of taking proof of concept to the next level if we're interested in spinning off a startup. But those have become increasingly more attractive over the years. And then the last thing is, oh, just for the future slides, you might want to add in respiratory therapists as your main, you know, with the doctors because I know at least one respiratory therapist that will be coming, they'll be like, hey, those doctors don't look at those at all. Right, thank you, thank you. In other nations, they don't have the title respiration therapist, so sometimes it matters. And I'm afraid SBIR's small business innovation research grants sadly are not given to nonprofits. You're not allowed to apply for them. So although my greatest dream is that someone will start a business based on technology made by public convention and make a million dollars. I hope that happens, I will help you make that happen. But I personally am not interested in becoming a vendor. Public convention is like pure open source. We want donations, we don't seek revenue per se. Yes, sir. So another important part that you talked about and I want to give some emphasis to is about around open standards. So the little chart that you have. Yes. If there are open standards around the data. Yes, it's called the public convention respiration data standard. There's a link to it in the talk. It's in a GitHub repo and it's versioned. Okay, because that's a spot where we're open hardware and proprietary hardware can come together. Precisely, that's precisely the case. Cover. But did you guys, you know, when I install Firefox or some other open source, you can, it defaults to telemetry, to send anonymized telemetry to some repository to collect the data from whatever I'm doing, like a web browser or whatever. For this sort of thing, if somebody is building it in the garage or whatever, I don't know, I know that's not the target, but I think that data would be very valuable to do, you know, kind of, I don't want to say clinical trial, but similar collect, basically use the data and then people can apply patches and things like that based on that data. Yes. So everything we've done so far, the Ventmon device, you enter a Wi-Fi password and it goes to a public data lake and this software is not as sophisticated as it should be. I'm working on this and this addresses what this reads, the data that you have there. But this, all of these things measured by IP address here are, in fact, this is a public data lake of these are actual breath traces and one problem I have is it doesn't move back and forward in time very well and a person can be on a ventilator for several weeks, you would like to have that. That's just, I need a software volunteer to work on that. But this software, when it's plugged into the Ventmon, it's live, it's not just looking there, it moves across the screen, it shows the pressure and flow the way a respiration therapist or clinician needs to see for a patient and the things on the right here are the PIP, the PEEP, the title volume, the respiration rate, the FIO2, the rise time and the fall time. Those are things which doctors are medically interested in and it computes it in real time. All this is open source JavaScript. Yes, sir. Yeah, can we get the mic back? Yeah, I was just wondering, are you based in LA? All this movement? Am I breaking a leg? No, are you based in LA? No, I'm in Austin, Texas. Okay, so let's say, I see a lot of hardware part to be done and personally, I'm very interested on the hardware side. I also do software, but if I want to really do bench test and all this development, I find very hard to find hacker space or work stations like some place where I can go and work on the hardware and test. Right. How can we deal with those kind of challenges if we don't have the right equipment? That's a good question. So the public convention policy, we call it the nuke policy, no out-of-pocket expenses. So I don't know where you can get a makerspace but what I tend to do is I buy inexpensive equipment for volunteers and you just do it at home, okay? Now, we don't have that much money. We get about $35,000 a year in donations which I expect to continue, but it's not enough to pay volunteers. I would like to pay people what they deserve but there's no way I could do, I would run out of money instantly if I did that. But I have found that engineers, if you buy them hardware things, they tend to think of it as like six times the value of what you're doing it because you're saving them out-of-pocket expenses. So the basic way we address this problem is if you need an oscilloscope, we'll buy you an oscilloscope. To do some of these hardware projects and to kind of have like a very basic electronics lab and the microcontrollers are cheap, it might be $200, $300, maybe $1,000. That's sort of a reasonable price because we can't pay people by the hour for our current model. You know. I have a question to my telemetry question before. It sounds like you have a GitHub repo for some parts of this project. I looked a little bit into this sort of technology also and there's already like an open source project called Oscar, I think, and it's more for the proprietary side of, I'm not sure, I think it's maybe for consumer devices, therapists I guess use this open source technology. I'm wondering if that's a way also to, I don't know, PR into it or somehow get more exposure to open vent or that sort of thing. I don't know if that's already been considered or not. Thank you. I'm not sure I'm familiar with that so I can't comment on it. There are some open source lung simulation software used to train people but it's a little different than what we're doing. But I don't know everything about the space. I may have missed something. Any other questions? So public convention started in 2018. We incorporated in 2019 before the pandemic. So we are devoted to open source hardware inventions in general, but the pandemic pushed us in a certain direction. So 70% of what we do now is related to open source medical devices. I have board members who want me to return to working on global warming. Their position is quite right. If we don't solve global warming, there's no point in making ventilator. But I have limited resources and I have limited intellectual resources, right? So I can only work on so many projects at a time. I wanna go back to the concept of the FDA approval as being important or not. Yes, sir. I almost think that I understand the reason to get FDA approval. It's a good baseline to make sure the devices are safe. But I also wonder if industry has deep lobby pockets to use FDA as kind of a gatekeeper. Maybe I'm wrong about this entirely, but is it something that should be pursued or not? If it's open source and there's already people who are in need of these devices that are just gonna build it with or without FDA, is it worth pursuing FDA participation? So let me express an opinion about that. I have seen corporations perform catch and kill operations. It has happened to me. I know that happens. I don't believe the FDA is corrupt. In fact, I believe a bigger problem is the fear that people like us will say, oh, the FDA is big and scary instead of just reading the regulations. Okay, if I have read the regulations, they're not that scary. You don't have to be a lawyer to understand them. If we do our paperwork, which I know computer hackers like me are not that great at doing our paperwork. But if we step up to that, I do not believe the FDA will be an impediment. There is nothing in the law or regulations which is counter to open source anything. People may think that, but it's often because they don't read the regulations about that. Really quickly, we have one last question, but I also want to say, Sean's talk in an hour and a half is going to be speaking to how to write your documentation in the collaborative way for open source hardware with a goal towards regulatory passing. Great. If you hold on to the FDA and once they're cool, that's what they have to call all their machines. Right. Well, so God bless you. I only have two comments on that. So first of all, it's important to understand the FDA doesn't actually test anything. The sense in which an FDA approved device is safe and secure is the following. They have forced you to do a lot of paperwork showing that you have thought about it. That I'm being completely serious, which is better than nothing. Okay. But it's not a guarantee that it's safe. Okay. The same problems that commercial firms have with safety, the open source community will have with safety. I'm sure, I hope 10 years from now, someone complains about an open source device being slightly unsafe because that will mean that we have open source medical devices. I believe in the end, open medical devices will be more trustworthy than closed commercial medical devices where you can't see what's going on. I think it will take maybe more than one decade to get there, but that is my belief. And if we have for five or 10 years, and then now they're actually done. Yes. Well, I mean, I believe we have to work within the legal structures and that means an open source firm may make the same mistake and they may be sued. And that's why these things have to be done by firms and for the next decade it's gonna be done by for-profit firms. And part of the reason those firms deserve a profit is because they have to pay lawyers to deal with that kind of liability risk, right? That's not something that fits in with the non-profit grant model. So what I'm suggesting is that people like me and possibly Darrell, maybe Sean, develop designs, but we don't become vendors. And then we make it possible for people who are driven to run businesses and want to deliver medical devices and take on all of the paperwork and liability and potential profit that comes with that that we support those people. Public invention is personally not ever going to become a vendor of medical devices. Well, thank you very much. Let's give one more round of applause for Robert. Thank you. And I believe we have an hour and a half break until Sean's talk. Maybe a little less now. Please contact me if you have any questions. In the talk, there are lots and lots of links about stuff that we've written and things like that. Thank you. You're welcome. Nido. Okay. Okay, thank you very much. Hey, thank you. I'll almost certainly talk to you about that. How's everyone doing? Okay, and everyone hearing me in the back? Raise your hand if you can't hear me. All right, I guess we'll get started. Sounds a little loud. I don't know how to turn this down. All right, so this talk is called Docs' Code Approach to Medical Open Source Hardware and PPEs. My name's Sean Marquez, as it says on the bottom. And I just want to give a quick thanks to the scale volunteer team. They've been putting a lot of work just to set up the AV equipment and networking and they put in a lot of effort to make these events possible. So I just want to give them a quick shout out. Hey Caleb, he's on the AV team, so. All right, so this is a quick line of my presentation just to give you an idea of whether you want to stay for this or not. We'll go over some quick turns. Like what is medical, what is a medical device or PPE as open source hardware? What is a Docs' Code Approach? Where are some of the stakeholder needs? Like why do we want to develop medical devices and PPEs as open source hardware? Let it alone using this Docs' Code Approach. What's the methodology? How do you actually adopt this approach to developing medical devices and PPEs? PPEs also stands for personal protective equipment, by the way, and an acronym. And finally, some community resources and places to contribute if you'd like to kind of foster this effort. All right, so first of all, what is open source medical hardware for PPEs? Quite simply, medical devices or personal protective equipment developed as open source hardware, which begs the question, what is open source hardware? Well, according to the Open Source Hardware Association definition, open source hardware developed is a hardware project that is developed and published as if it were open source software, right? That's the very basic definition. A little more on the legal side, it should include a license such as the CERN Open Source Hardware License, which allows for modification and redistributions of the hardware design. Again, that's as per the Oshawa definition. And the documentation and design tools themselves should also be open source, right? This is from the OSI or Open Source Initiatives definition of open source software, which vicariously also applies to open source hardware. So what is Docs as Code? Has anyone heard that phrase? Docs, we've got one, two. What about docs? Has anyone heard of docs? Documentation, how about code? Okay, we've got more hands. All right, so Docs as Code is basically the philosophy or practice that documentation should be maintained using the same tools and approaches as code, right? And we'll get into that a little bit later. But first, why would we want to develop medical devices or PPEs as open source hardware literally alone using this approach? So one, I think, is distributed manufacturing. So Robert earlier kind of talked about this a little bit, but when you have vulnerable abilities in the supply chain, you don't have quite the vendors quite at your disposal. Maybe you're in a third world country and you don't have the shipment supplies coming in to be able to build the devices you need in time. This might be of interest like something maybe during the event of a global pandemic. I don't know if that's ever happened. And Robert also mentioned this last talk, but I think it also yields just better quality products, right, kind of by Linus's law, the phrase given enough eyeballs, all bugs are shallow. So if you have more people looking at your projects, like more likely they're able to kind of find something that's wrong with it and then maybe contribute, having breathing problems and realize they have to still have my mask on. All right, so why use the Docs as code approach? One is that it works with modern version control tools like Git, right, because documentation is plain text that it's compatible with modern version control tools, which by extension, if you're using a distributed version control system, it also allows for parallel development, right? Reviewing documentation becomes like a code review, right? So if you're hosting your documentation on something like GitHub or GitLab, reviewing the content is as simple as opening up a pull request. By extension, content can also be open source, which means you essentially open up to the entire world to be able to look at your code or documentation, which essentially is outsourcing your entire documentation to the world, say, hey, look at my Docs, right? And by extension, a lot of tools have been developed that utilize this approach. So there are a lot of Doc tools that will work with something like a markup language to be able to produce documentations for generating something like a PDF or HTML document that might have the interest to specific stakeholders, right? And I didn't quite add it in my presentation, but another methodology that you can, for advanced use cases, apply is a model-based, what's called a model-based approach to documentation. So that essentially means the contents of your documentation can actually be stored in a very structured data model, which is machine-creable, which is pretty amazing, because then you can essentially query whatever content you want out of the documentation and any relationships between model elements that pertain to maybe of interest to specific stakeholders, right? But that is a much more complicated talk that I'll probably say for those future talks, right? And speaking to that use case, like that might also be of interest to the regulatory environment, such as the FDA, or if you're trying to get like a EOA certified, if your documentation is query-able in such a way that you can produce very specific artifacts that are of interest to specific stakeholders, then you can essentially show the regulatory environment and agencies that you've developed processes and methodologies that conform to whatever specific regulatory environment, right? All right, so how do you actually adopt the DOCSIS code approach? I would argue you need three things. One is a language, a tool, and a methodology. So where are, let's go through each of these items. So a language, like we mentioned, you might wanna use something like a markup language, like markdown, restructure text. If you work in academia, you may have heard of LaTeX. ASCII Doc is another one. These are all markup languages that are human-readable and plain text, right? Which means that it's easily compatible with modern version control systems like Git. Another one, which may not be as familiar as a 10-footing language. I like to think of these as languages that help you transpile one language into another. So maybe you have content that's in JSON and you wanna reproduce that as YAML or something that can be generated as a diagram, like plant UML. You can actually write templates in these things called templating languages, such as liquid, ginger, handlebars, to name a few, that actually let you transpile one language into another, which I think is kind of cool, right? And also templating languages are used for formatting. So if you're producing something like a HTML document, you might want additional decorators that allow you to add styling and formatting to a artifact that you're producing, such as a webpage, right? And the third one that's not utilized all that much, but I think is starting to kind of make itself known are modeling languages. So modeling languages are usually used in aerospace industry, but I think they're starting to kind of pop out in the mainstream technical writing community. Just curious, who here has heard of a modeling language? Oh, hey, quite a few, cool. So I listed two here. One is OML, which is short for the ontological modeling language. So that's a plain text modeling language that's being developed by JPL. SystemL2 is another one. So that's being developed by a bunch of a working group in like mostly the aerospace industry. Basically, what it means is you can define your documentation as a model, as opposed to a document. Well, what is the difference, right? The main core difference, I think, to keep in mind is that a model is machine-creable, whereas a document is not. So if you produce something like an HTML or PDF document, unless you wanna pull your hair out, for the most part, it is not machine-creable, whereas a model is. And depending on the complexity of the model, you can add a little bit more structure and semantics to give more relational meaning. So maybe you want to model something that has a composite or an aggregate relationship. You can get very abstract and complex with modeling languages that you can't otherwise use with these other languages. All right, so you got languages. Now you got tool chains, right? So some of the tool chains you'll need to apply this approach is a text editor. I listed a couple here. So I personally like Vim. You could use nano or VS code, but basically something that can mutate plain text. You wanna use a version control system like Git, SVN, Mercurial, and then a static site generator or rendering engine that can read these plain text languages and then convert them into something a little bit more familiar like a webpage or PDF document, right? And then this is mostly for the project management side, but if you use a platform or issue trackers, then you can manage issues around your documentation on platforms like Jira or GitHub or GitHub or what have you. And then of course a publishing platform if you wanna make your documentation public. So some popular ones are GitHub Pages, read the docs, Netlify, and then if you wanna get a little bit fancier, you can automate some of these production pipelines so that whatever you publish commits to your repo, you can automatically generate a build that publishes directly to your publishing platform of choice, right? So GitHub Actions is a built-in CI pipeline and GitHub and then Jenkins is another one. And then last but not least, you need a methodology. So where are some methodologies? So one such methodology may be called Docs Driven Development. So this idea that you write the docs first and then implement what you documented. So this ensures that your documentation which essentially serves as an API for humans conforms to what you expect your project to do, right? Another thing is the contributing guidelines. So style guides, code of conducts, like how do you expect your documentation to look if you're soliciting the help of contributors from the web, right? And just general behaviors for how you should behave in the community. If you worked in most software projects, you've probably heard of Agile. So you can apply a practice like Scrum or Combon for deciding how do you wanna release your documentation, and because your docs are managed the same way as code, you can have doc reviews, right? So we mentioned pull requests having, I don't know, does everyone do code reviews? Does anyone? Okay, we have, oh, you can do the same thing for docs. So some community resources, if you wanna take this approach on, write the docs is a community of technical writers that have essentially their own online Slack channel and talk directly to the developers that develop these doc tools. Maybe you have a feature that you really need for a use case or maybe you just wanna file an issue that's not getting enough detention. You can often just go to the Slack channel and then talk to these guys directly, which is kinda cool. They also have a podcast. So if you're stuck in LA traffic and looking for a new podcast, there you go. The Open Source Hardware Association also certifies open source hardware projects. They also have a Discord channel. And as of recently, they also have a standards working group, right? So they're trying to standardize or develop open standards, very similar to ISO, if you've heard of ISO, that open source hardware projects or companies can kinda just adopt and implement either tooling around, which I'll talk about a little bit later. And oh no, his icon didn't show up. But the Mock30 Foundation is a nonprofit originally chartered to develop open space hardware, which I think it's kinda cool, but they have recently been focused more on methodology and processes and tooling, just because of complications with ITAR and whatnot. But they specialize in a model-based approach to documentation, which I spoke to a little bit earlier. TetraBioDistribute, which icon is also not, oh, you know what, because my wifi isn't working right now. So that's why my build for this particular icon isn't showing up. They are a nonprofit pioneering the development of medical open source hardware and PPEs in collaboration with a lot of subject matters, like at USC, like Daryl. And if you wanna contribute to any of these projects, these are some of the ones I listed personally. So the OSS, the open source hardware standards group is developing a lot of these standards that I mentioned specifically for configuration management and quality management for open source hardware for industries such as medical and aerospace. The Distribute Source Hardware Framework is a framework kinda like, I kinda think of it as the Node.js or MPM package management for Distribute Open Source Hardware. So how do you modularize these open source hardware projects such that you can reuse it in another project or utilize something that someone else has already published. And part of building open source hardware is that you also need the tools. So they're doing a lot of work around not just telling you the bill of materials, assembly instructions, and supporting documentation you need to build the thing, but what other tools do you need to build the thing, which may also be an open source hardware project that has its own bill of materials, assembly instructions, and supporting documentation. So it gets very recursive. And if you wanna contribute more on the tooling side, TetraBio is also developing a MVP command line tool called QMS CLI. So that's kinda like a documentation scaffolding generator for doing quality management around open source hardware that needs to conform to these regulatory environments. One of the flagship projects that Tetra is working on is called the PAPRA. So this is an open source hardware respirator. So if you wanna work on something more tangible and physical that you can source and build yourself for less than 300 bucks, okay, 150 bucks. You can check out the PAPRA repo. And this presentation itself is also published using a DOCSIS code approach. So if you wanna improve this very presentation, you can go to this repo. So that's my presentation. Let's go build some mobile source hardware. One of the things, so my name's Mark, I'm running the track, but I also work really closely with Sean and one of the things I really wanted to impress on you guys is that every in our journey, like we didn't know anything about building hardware devices, especially medical hardware devices when we started doing this. And we met every single time we wanted to do something cool, we would run into this or that regulatory body that would tell us what we needed to make sure that we had. And so a lot of the work that Sean has done has made it so that every time we make a documentation change or a design update or anything like that, it gets tracked. And so those design updates, they have to be tracked as far as the government is concerned, as far as the regulatory bodies are concerned in order to understand your thought processes as you are making the object, right? So there's a lot that goes into that as well as what are your testing procedures. All of that stuff has to be tracked. And what makes this tricky in hardware as opposed to software is the logistical management of getting those devices to everybody in order to make the docs. So one of the things that I reason I asked Sean to give this talk is it's like in this environment, in the hardware environment, you are targeting the document that the regulatory body wants you to write. That's your end goal. You have to say, my device meets these things. And so much the way that we do BDD, your behavior driven development, right? Where you say like my code has to do these things, you're saying my device has to meet this document. So you write the document that says what my device has to do and then you work backwards from that, right? And then that's what this framework allows us all to do. So I want to make sure that we all, that that message became abundantly clear because that's when we talk to the FDA, we talk to NIOSH, we talk to the CDC, that's what they want to see. They want to see these docs showing that we know what we're talking about. So, any questions? Is there quality control? So, do you want to answer? Is there quality control? Not on the appra at the moment, but that's one of our stretch goals is to like be able to develop standards and tooling that like you can publish a project to be able to conform to. Right, so in the docs as code approach, as that quality problem arises, you submit a bug in like a git bug, that gets logged and then that way you track it and that's what these regulatory bodies want to see. That you've seen that there's an issue and that you're tracking it and that you have a pull request against that issue. Like the way that git flow works, like they don't even like when I started talking to these guys about here's what git does and their responses, I have no idea what that is. All I want to know is that you have work that goes up against an issue. Like, well, that's what we do. So, any other questions? All right, well, one more time, thank you, Sean. We'll be back here in this room in about 30 minutes or so. To, yeah, in about 30 minutes to go over the appra, which is the project that he was talking about at the end of his presentation there, which is an open source power respirator. So, which we will have actual working copies of up to show off. One, two. Da, da. Now, now, now, now, now, now, now, now. Testing, testing, one, two, three, testing, testing. One, two, do, do. Does this one work? This one works. Testing, one, this one works. Maybe we can start karaoke early. I'll use this unless they get this fixed and then I'll go back to the boy band mic. Is it Britney Spears that likes this mic? Somebody, somebody that performs likes this mic. Other side, okay, yeah. It's gonna bring my toxic outfit too. I've had many requests not to show up in that. It's exactly three o'clock. I'll give it a couple more minutes. Okay, he's running, let him do it. Feels really weird to be indoors without a mask. Actually, I've been told this helps a lot more with my photographs. Early on in the pandemic, I was at a total wine and I was wearing a 3D printed mask that we were working on and a friend from high school that hadn't seen for like 15 years or so stopped me in line and it's like, hey, Daryl, is that you? And I was like, what, how did you recognize me? It's like, oh, I saw your posting. Sorry, we had technical difficulties. Gonna try to fix those. Can I start with this one or just wait? Oh, I haven't started to talk yet. I'm just waiting for you to show. Yeah, just, oh, okay. Okay. So today I'm gonna be talking about the PAPRA, which is an open sourced powered respirator in response to the COVID-19 pandemic. So give you a little bit of history. First of all, let me introduce myself. My name is Daryl Huang. I'm an assistant professor of research at the University of Southern California's Keck School of Medicine, working in the Department of Radiology. And I'm here with TetraBio distributed, where a nonprofit set up for open source hardware designed for medical applications. So just in case anybody hasn't heard about this thing called COVID-19, just to give you a little bit of, it is short for coronavirus disease, 2019, not the iteration like a scale of the 19th version. It was not the 19th version of the coronavirus disease. It is caused by the SARS-CoV-2 virus. You might have heard of the SARS-CoV-1 virus, which was actually SARS. And we can talk about, and why did it spread so far and so quickly? Number one, infectious people were infectious days before they presented symptoms. And so people are unwittingly infecting other people because they didn't know they were sick. Also, this is a novel virus, meaning that we didn't really have treatment or understanding. And finally, we don't know the exact spread mechanism. Was this primary surface contact? Was this airborne? Turns out it's a combination of all the above. It's actually a little bit more difficult to spread than worst case scenario, but some of the newest variants that have come out have been shown to be even more infectious. So that might not even be holding true these days. BA5 apparently is as infectious as measles. So joy. It's not over yet. No, okay. So just to give an idea how this has affected us here just in the United States, this is a month by month chart put out by the CDC of COVID deaths, just deaths. We're not even talking about the hospitalizations and the people that are suffering from long COVID or recovered, this is just deaths. And to summarize this, this is approximately that many people in the United States only that have passed away from COVID or at least being COVID positive. Even if you're at the most conservative estimates, this many people dying is not normal. So you might have heard some statistics out there. It's like, oh, this isn't as just like any other cold. Nah, it isn't, it's not even close. So why do we get onto this entire path of looking at a PAPR? And that's because in March of 2020, this was the actual guideline that came out of the CDC. They told us in healthcare that if we ran out of masks to wear a bandana, okay. Yeah, we were in that situation. New York, LA actually was fairly insulated from that. We were lagging behind New York in terms of cases. New York definitely hit this where there were healthcare providers, frontline healthcare providers that had no PPE and were making do with bandanas and scarves around their faces while treating active COVID patients. And so just to give you a little timeline of how this project started, we started out at Keck with a appeal from my clinical director from the radiology. He said, hey, we have this 3D, I saw this 3D printing thing over on the East Coast. Can we do it? And in the course of a couple of days, we basically ramped up, talked to a whole bunch of people and by the end of the pandemic started doing things and started making things like 3D printed respirators and face shields. In fact, our face shields were actually clinically deployed and we gave out, I think close to 6,000 of them just within our organization. And mainly because the reason we didn't give out more is because we didn't need to. We actually gave one to every person that needed it and they were cleaning it and reusing it. We printed about 9,000 of the 3D printed masks that were never deployed because we actually in LA had enough PPE through beg borrowing and stealing all across everywhere. We were able to keep enough supply of good N95 masks for the people that needed it. And actually our next talk is going to be talking a little bit about that, the entire shortage of N95s and the issues around that. But that's what it looks like. There's a preview there. Huh? There's no preview. Nope, always a preview, always a preview. Yeah, sorry, a spoiler alert, yeah. So going on from there, this is the picture actually that I was telling you earlier that my high school friend recognized me off of I hadn't seen in 15 years because I was wearing this mask while shopping in total wine and just stopped me. Anyways, we basically ran a community printing effort and this is actually a part of the genesis of the ideas that we had for open hardware design. We actually reached out to the community and went to maker spaces, talked to people in the community that just happened to have 3D printers. And then we tapped the large untapped resource which was our School of Architecture had like a roster of over 200 students that actually just had 3D printers at home as part of their normal usage. And we started ramping up and doing 3D printing in a community sense. This is just pictures of the stuff that we're making at the time. And then, simultaneous to that, we were talking about what else can we do? And we started working on the PAPR. We started off with this organization that was known as PAPR Life Force. And like many things that started during the pandemic, it was well-intentioned, it grew very quickly and then it flamed out. It dissolved in September. And that's actually when I took the PAPR and brought it to TetraBioDistributed which was another organization I was working with on a different project which was a VentSplitter project. So Tetra started off its life as a VentSplitter project working in March 2020. It basically consists of a very diverse group of engineers and scientists all around the globe. We had people in Colorado. We had people in England and Great Britain in general. Yeah, so it was a very widespread and through the magic of the internet and the fact that none of us were going anywhere outside of our immediate dwellings, we had a very intense group working on projects during the beginning and middle of COVID pandemic, well, beginning and middle of the COVID pandemic as it was progressing. And yeah, VentSplitters are hard. Well, we can talk about that on the side even if it wasn't a pandemic. Yeah, because frankly, they're kind of a weird middle device that's very difficult to get right. But TetraBioD, we kind of pivoted towards the Fitra which is actually something preview spoiler Daniel's gonna be talking about. And the PAPR, which are more achievable goals. And why do we say that? It's because these are two devices that are used as PPE as opposed to devices geared towards direct medical intervention. So a splitter would be, you're ventilating multiple people with one ventilator. That has an entire set of regulatory issues associated with this. PPE has a no less complicated set of regulatory issues but a different regulatory issue pathways. And then so at our peak, we had about 30 volunteers and now we have about 10 that work in the course of this. And again, it's like many of the pandemic groups, people find jobs, people start working on other projects. And to find more about this, please check out these two websites. The first one is our standard website and the other one is our GitHub, which actually you can find this entire project there. So let's talk a little bit about our design goals. So these are some of the design goals that we came up with right in the beginning. And we wanted something that provided at least N95 levels of protection because anything less than N95 it wouldn't be useful for us. Initially we're thinking about this in the context of hospital care or frontline workers interacting with other people and with possibly COVID patients. And so N95 is the bare minimum level that we wanted to reach. Also we wanted something that could be worn for a long period of time. These disposable N95 masks are only rated for four hours at a time. So a well-fitted N95 mask, you're only supposed to wear it four hours at a time. That doesn't mean we don't wear it for longer, but it is only supposed to be worn because after that amount of time there is an increased resistance that you have to breathe through. So you might actually have a little bit of effects, headaches and stuff like that, if you wear them too long. And that might be inconvenient for somebody, but if you're in a healthcare situation and you're making life or death decisions that might actually become an issue later on. Also the entire comfort aspect is you might think, oh well, if it's just annoying on the face, that's not that bad. Well, when you do an eight hour shift and you're getting pressure sores on your face because these masks are biting into you, that becomes an issue after a while. One of the other, the caveat says we needed to be able to change the battery very quickly because you don't really want to stop breathing in the middle of a shift. It needs to be lightweight. This wouldn't be useful if we're carrying around a cart with it. And in 2020 we wanted to use non-COVID rationed components. That means something that wasn't already being provided to the hospital. So for example, if we could create a PAPR that used PAPR filters and we couldn't get PAPR filters, that was a pointless device. Attachable to a belt backpack caring system. Most PAPR devices are actually a belt system already. And we wanted to mimic that and be able to modify off of that. Easy donning and doffing. Donning is putting something on. Doffing is taking it off for those of you that have never heard of those terms. The reason why that's important is across contaminations, one of the biggest ways that you can actually spread any kind of pathogen. It's you contaminate your hands and then you rub your eyes. That's actually why we tell you to wash your hands so often during this pandemic. And finally, 100% open source where possible. Because what we see is that places, let's be honest, during COVID, one of the issues that we found is that supply chain, when it breaks, you can't get anything. And if you can't get it, no matter how much money you have, it doesn't really matter because you can't get it. It just doesn't exist. But beyond that, one of the problems is that we live in a world where healthcare inequities exist. And a PAPR unit generally costs around $1,600 on the low end. And there are plenty of places around the world, even within our communities, that don't have the resources to outfit everybody that needs a PAPR because of sheer cost. And so we wanted to keep the open source aspect so that people could continue developing and also using this without paying exorbitant amounts. And we wanted to design this for distributed manufacturing. Again, going back to the shortages, one of the problems that we had was the fact that our supply chains were cut. So traditional manufacturing, which most of it is done overseas for these devices were not available to us. And looking forward, we would like to have something that is able to be essentially produced locally on site where things are needed, as opposed to far, far away and then shipped to a place. Because you never know what the next pandemic is gonna be. You don't know what the next outbreak is going to be and where a lot of these devices will be needed. And finally, the last design goal that we came up with because we talked to our UK and people that live in areas that apparently have this thing that comes from the sky in terms of water. I think they call it rain, splash resistance because a wet filter is a non-working filter. Oh, and since then, one of the entire rationed components we've kind of abandoned because we don't need to live up to that anymore right now. Here's the example of early prototype. Number one, it never worked. You probably can see the errors in this right now. The fan was outside of the sealed box. So it was sucking in slight amounts of air which is just enough that it would fail the tests that were useful for it. But it was useful in looking at our 3D printing would work as a technology to create components. And it was sort of the genesis of the design aspects of this. Here's our October 2020 iteration. Going forward, all these iterations from now on actually work. They actually passed N95 rating on the test. What we've been doing is creating basically a more robust PAPR unit because a lot of times these were very finicky to get too past, you might have to tinker around with it. And sometimes it didn't work for other reasons. For example, this one, we decided to try a dual hose method as you see Mark there wearing the dual hose mask where the inspiration and the expiration goes out the same filter. And that turned out to be a bad idea because we did a CO2, entitled CO2 check and found out that it was a great way to rebreathe there. Wasn't deadly unless you kept on going for hours. But all the, in this case, all the components were sourced from Amazon. The next level, so starting in 2021, January, we tested out some new masks. So the other aspect of this is we're trying to make 3D printed masks that would fit on the face that you could actually make yourself. So we actually looked into printing in flexible materials like Ninja Flex. It's an incredibly challenging field of printing flexibles, which we did not fully comprehend in the beginning, but we quickly realized that it is not simple. 3D printing isn't really simple to begin with, but that was even worse. Yeah. Sure. So eventually we ended up with that type of design. The inhale and exhale valve would have been fine, except for the fact that what was happening was the exhaled arrow was just getting sucked right back into the filter. So that was the problem with using one filter. Yeah. So that's why we got rid of that particular version. But as you can see, we're trying to iterate on this and we're learning a lot of things and finding out where the issues were. And yeah, learning process. This is actually the first model that we created a board for the power and we'll talk about that as we get into the design. Next iteration, as you can see, it's about every month or so that we're doing these build parties. We made some board changes and then we made a bunch of masks that felt okay, but we're still iterating on design. And then we got into quantitative testing. So just very briefly, how do we test these things? That's actually a very good question because at the beginning of the pandemic, the standard way to test an N95 mask was a qualitative test. You put one on, you throw a hood over somebody, you spray chemicals in there, either Bittrex or saccharin, and then you say, hey, do you taste anything? Right? Big problem. What's one of the symptoms of COVID-19? Yeah, that's one of the problems, right? Well, going along that, we decided that we needed a more quantitative test. So we actually gone to actual particle filtration testing, which is the quantitative way that you would do this type of testing, but we've adapted the rig so that we can test our devices also. So this device that you see here is actually a port account. It's normally used to test these disposable masks for fit, but we've basically made it so that we can test our devices to see if it's filtering at least at the rate that we're talking about. And it's important that we were doing this because the filters that we were using at the time were just standard HEPA filters used for air purifiers. And we also learned some things along the way, among which is if it says HEPA, true HEPA, or any of those variants of true HEPA, those actually filter at the N95 level. If it says HEPA type or HEPA-like or any of those other variants that are not true HEPA, those filter at 10 times worse than the HEPA filters. And that was actually a very big revelation because none of this is really documented quite clearly anywhere. So when you go home and check your air filters now, you might be like, hey, wait a second, this thing is doing nothing. Okay, so yes, as I said, the TSI port account. And then this one we experimented with a proprietary connector because we're looking for a locking connector for power. Anybody here do hardware? Yeah, do you know of a good connector for power that locks and is water tight? Yeah, yeah, that's one of the, or at least, you know, splash resistant. That's the problem we came up with is that look, these connectors would cost as much as the box, so yeah. So this is April iteration. We started, we changed it up, we moved all the inputs onto one side of the fan box. Therefore, the entire fan and funnel unit that you see there with the output, that was actually one complete unit when we completed it. And then you just slotted it in. We also started working with some more moisture protection. We just made louvers so that the water would sluff off a little bit. And then we added BNC connectors. We're actually using BNC connectors as our connectors so that they don't disconnect. The important thing here is that we have a power supply that is not firmly attached to our power unit. And what happens when you use just a regular, you know, 5.5 millimeter jack? Jack, well, you move a little bit and it pops off and suddenly you don't have air being pumped to you. Which is something that most people don't like is suddenly having their air flow stopped. So we had to redesign a little bit. So the May 2021 iteration now with the actual physical weather guard on top, this again was to help out our people that are living in rainier climbs, mainly just to prevent water from getting directly onto the filter. Because if you saturate a filter, that becomes essentially a wet cloth. And a wet cloth over your face is called waterboarding. And you don't want to do that to yourself. And this is one of the reasons why we're looking at this. Now we tried an iteration where we 3D printed all the 3D printed clips to hold everything together because we're trying to get less in the number of parts that didn't work very well. So we went back to our tried and true screws. And then we started testing flow testing and battery. So one of the things is that, and it's kind of hard to do is correct testing. Like trying to figure out exact flow of air through a circuit is not the simplest of tasks. And there's also a lot of conflicting guidelines on how much air that you need flowing. The standards are actually very vague because some of them there aren't true standards. They're actually just guidelines that are posted. So we've been working our way through that trying to balance the power consumption with the type of fan that we have and different grades of fans. One of the things that we've been trying to do is use fans that are made of traditional blower fans, so 12 volt blower fans, because you can actually access those. There are specialty fans made for CPAP companies and those can range over $200 a fan. And we're not looking into those because the average person's not gonna be able to get that. So we're looking at devices that you can order off a Digi-Key or Mauser or Amazon. So that's one of the design constraints that we have. Yeah. Was it 150 LPM? No. Well, so there's actually a couple of reasons. Traditional pappers are actually vented into the open space. And so you need enough air flowing so that there is no chance of taking a deep breath and back breathing in. Plus you have to fill that entire space of the hood. So that's generally the six CFM minimum that you have set. In fact, that actually goes higher than that. Most pappers, I think, get up to 12. I think they can push 12. But those are all hooded pappers. Half-face refrigerator pappers are a little bit more uncommon and they're kind of a sub-variant of that. And that's why I think it was four, two to four, right? Is the, that's four to six. The question is, is whether or not you'll actually ever use that? But let's wait until the question answer because, no, no, I have more of a resident expert in that side that can talk about that. Yeah, yeah, it works both ways. Okay, now in February 2022, yeah, we started live streaming because we were actually showing people how we build these units. So in addition to having all our files online in GitHub that you can download, print it up and the circuit board and the circuit diagrams online and being able to fab a board if you want, we also went the extra step of showing people how to assemble from scratch all one of our units. And we've been doing that. So in June, 2022, we did another one and it's now on YouTube and it's also on Twitch. So if you're interested in checking that out, these slides will all be made available on the website, right? So that you can get that or if you wanna take a picture. Yeah, yeah, hopefully you can download the slide and it'll be a lot easier. So give me an idea, our current generation, this is the latest one that we've developed is much smaller. I actually have some here so we can talk about it afterwards but if you take a look at it, it's basically a fan unit. We're actually using two P100 filters. These are the client tools or GSB, half face respirator filters and we're using those as the filtering material for this. Some of you have probably seen the 3M filters, right? Like yeah, the 3M filters. One of the problems actually with 3M filters, they have a proprietary bayonet connector that's really annoying to actually 3D print something that attaches cleanly to that and it's a very small opening. So we wanted something a little bit more so we've been experimenting with other filters. The original filter that we started with actually was a germ guardian air purifying filter. Yeah, yeah, a germ guardian filter which is a larger format and this is one of the reasons why the format that appears bigger than the other boxes is that we just change filters. So just to break it down, so this is actually all the components of a fan box. Everything here is essentially you can make it except for these two pieces and the screws. Well, I guess you can make screws if you're really good but we normally order the screws from McMaster right now and then the fan and the BNC connector are the only two components that are not custom built. Everything else, including for example, right here, this is the weather guard and all the plastic pieces are all made for 3D printing. We've been printing this on a Prusa and I've actually printed it previously on an Ender 3. We printed out a PET-G which is a little bit more resilient than PLA but in general, one thing to remember and it's been brought up a lot which is oh, 3D printing, it's permeable, it has, you might have issues with it. In essence, all you have to do is prevent air from passing, right? You're not gonna submerge this for hours on end. You're not going to try to hold liquid in it. You're just trying to prevent air, you're trying to channel air through a filtration device and if your filtration medium like the filters is more resistive than the plastic, then you probably have other issues that you're working with. But in this case, what we have is the, everything is 3D printable and we've optimized it to print on FDM printers. So everything that we do here is on FDM printers because most people don't have resin or any other powder technology at home. The other thing is that all our seals are actually made out of craft foam cut on a Cricut. So we have Cricut files and cut it on craft foam. It's actually really, really useful for those of you that need to make custom seals for boxes. It's actually a quick and easy way to make seals. Highly recommended if you're doing dev work where you need seals. So going to the power supply, we wanted something that is interchangeable, right? So that had a charging unit that made it easy to work with. So we came up on using power tool batteries because A, they're plentiful. In fact, I believe Milwaukee is the largest importer of lithium ion batteries in the United States because, well, at least the parent company, they also make Ryobi. So power tools, lots of them out there. And then also they're ruggedized. They're created for use. And so we adapted to that and we created our control unit to use one of those as the power supply. What's important for us is that we didn't want to have to develop an entire battery system and you need something interchangeable because in the middle of use you might need more power. So integrated battery is a bad idea if you need to take off your unit and charge it in the sod. Here's a very quick schematic of the different components. Both of the housing is 3D printed. Also, it is using PETG. The board, we custom fabbed a board. Patrick designed it and we'll take a look at that in a second. You can see the battery. And so yeah, this is the PCB. And as you can see, we have many different features in this. Going off the top, we have connectors for the battery. There is the potentiometer that switches it on and also controls the speed. So the speed of the fan, we can adjust it from very, very slow to very, very quick. We don't have steps on it right now. It's just a swiping flow, so you can adjust it more or less in an analog way. The battery has a gauge, the LEDs. There's four LEDs that signify the battery level. It's important to give a visual and we also have a buzzer in there that gives an audio sound for when the battery gets low. One of the things that all PAPRs require is a visual indication and the visual indicator is not within the field of view. I believe it needs also definitely an audio indicator that you don't have enough power. So yeah, we have a PWM. Now it's just enough we added in a power jack. So there's a 12 volt power jack on the inside of the connector. The reason why that's actually kind of really nice is because if you're in a car, you can plug that into your car and it'll run off of that. So we thought about our Uber drivers and delivery people in terms of usage. And as you can see, most of the other little portions. And overall, in quantities of 100, they become fairly affordable. So the entire cost $13 for in quantities of 100. And here's a good look at the board. If you have any questions at all, please come up to Patrick and talk to Patrick. Yes, there's this bus that I'm throwing you under. Oh, and we also included an eFuse in there just in case you trigger by, let's say, plugging in that 12 volt into the wrong plug or backwards or something. Okay, the last component to this is the mask. And what we've done so far is we've created these adapters for the 3M half-face respirators. Going back to that entire annoying bayonet socket that they have, we've actually created 3D printable models that allow you to attach this to a hose setup. And forgive me for using a picture of myself, this is like. So you can get your own main mask set up that essentially will give you filtered air. But it all stays nicely contained. And the actual 3M face masks are, if you get the silicone ones, actually are very comfortable and they fit quite well. We have the physical models, which we can, if you wanna come up and look at it closer. Quick acknowledgement of the team. These are members of our team. We wouldn't be here without everybody on this team because as you can probably tell, this takes a large team, not just a single, no single person is going to be able to hack all the way through this. And this is a cross country, cross national team. Jamie right there is actually over in, I believe sunny right now, UK. It's actually quite sunny. And Burhan and Kevin are both on the East Coast. Mark, Daniel, me, Patrick and Sean, we're all here. So if you wanna talk to us, we'll be happy to talk. So what's our next steps? Solutions for other than COVID. Well, Papas are really useful, especially in California where we catch on fire every other week and air quality suddenly goes from eh to ugh. And you know, Papas are useful. We're trying to solve the proprietary mask issue because getting a good mask on the face, 3D printing right now probably can't handle this. So we're looking into ways of solving that. We are trying to publish our current iterations so that more people can it. And you know, we might go for a Kickstarter. Anybody have an extra $2 million that they would like to donate? Mark is right here. He'll take the check. But for call to action, what we're looking for is we need help. We want help getting the word out. We want help if you have any design chops for getting a mask solution. Or if you know anybody that wants to work on mask designs. And help building. So we want to make sure that this is something that is doable from people just looking at our website and just printing in themselves at home. So if you have a 3D printer or you know people that want to take on a project like this, a maker space in your local area that wants a project, direct them to our website and you know, we'll be happy to interface with them if they need help. But hopefully everything's already documented and they can just go to the website and just make it themselves. And then finally documentation and instructions. Most of us are engineers and I believe literally challenged. Or verbally challenged actually. Literally challenged, but actually just documentation and also converting instructions to a wider audience is actually pretty important. Yeah, different languages also. Yeah, as Mark was saying before, we've actually have members that were in other countries. North Africa was one of the places and the different language aspects, definitely a challenge and we'll need if we want to get to these other markets and help some of the places that could really benefit from this type of open medical hardware, we will definitely need language help. And yeah, that's my presentation. Thank you. I think we have a question here. I was going to say why don't you repeat it? Ah, okay. So the question is about regulatory filings if we've made any regulatory filings. Right now we haven't. We're not currently going through the NIOSH process. So PAPRs are actually governed by NIOSH government agency that certifies these PPE devices. We are currently working on making sure that our devices, A, can be made, right? And then B, past the bare minimum testing that we're putting through it right now. We're hoping to get it to the level where it could pass NIOSH, but one of the issues with NIOSH certification is that it is not designed for doing yourself build it at home. NIOSH certification requires you to certify the entire chain of creation and manufacturing. So an open source device will probably never pass the current set of NIOSH rules because they have to go and, for example, you need an ISO 9000 certified factory that is governed by NIOSH. And that's not gonna happen with an open source device that you're making for at home. Yeah, yeah. Well, so FDA doesn't actually regulate that space. This is, NIOSH regulates the space. So NIOSH could come up. So the reason why you want NIOSH approval is because if you wanna sell it commercially and you want it to be used in regulated spaces because NIOSH is closely associated with OSHA. If you're in a field that requires the usage of a PAPR it has to be certified by NIOSH. Otherwise, you're not covered under your protection of your employees and therefore there's liability issues there. So if you're gonna be using it on a commercial venture or employment as a protective measure, then it's probably not gonna happen with an open source in the current regulatory environment. Keep in mind, the pandemic's really shifted a lot of thought processes on this because one of the things I pointed out in the beginning, they told us to use bandanas, right? They told us to use bandanas. At the beginning there was a lot of emergency use authorization for a lot of things. If we get into another situation where we're in a pandemic situation, we might hit another time when they're basically saying use what you got. What we're trying to do is give you, in this project, to give you a reasonable alternative that's better than a bandana, right? Yeah, actually, so you had a question next and then there was, do we answer your question? Oh, okay. Excellent question, so the question is, is that we are testing it with equipment that most people don't have. How do we guarantee that if you do it at home that you're meeting to the same standards? And that's actually open to debate right now. We will be in our build guides, give you best practices, right? Of what we found that works. Unfortunately, like any kind of implementation at home for anything that you do, it's subject to the skill of the person that's doing it, right? And there's no real way to guarantee that. And that's actually one of the problems with trying to go with NIOSH is that NIOSH can't say, well, this person knows how to run a 3D printer. That is a problem, not yet. But that's something we can look into. Right now, we're trying to get devices that seal, what we do is we try to build in the design enough overlap that there shouldn't be a failure with the tolerances of the average 3D printer. But for example, if you start printing and you have gaps in your 3D printing where there's air leaks because your printing isn't good, there's nothing we can do to control that right now. I mean, that's a valid question of what do we do about that? And it's actually a problem in FDA certification for devices is what do they, how do you regulate that space? Right, so what Daniel was saying for just to record is that we need to figure out ways to do rudimentary testing for quality control in a decentralized way. And that's an open problem still, right? So that's something that we definitely need brainpower on. In the first presentation of the series in the morning, we talked about testing and the need for independent third-party testing. And one of the issues with PPE is that a lot of that testing equipment is fairly high-end and it's not generally used in other fields besides the particle testing, right? You're looking at particle filtration. So there might be only select centers that can do that just because the equipment doesn't exist as a, it's not as common as an oscilloscope. So you're gonna have issues with that. Again, these are still open issues, but I like to go back to the original thought process behind this entire project, right? Which is my mantra in the beginning was do better than a bandana. And it is a sad statement that in 2020 that that was my mantra, but that was my mantra. And this is the progression of that work, basically. Yes. Yes, let me go back to. Are you trying to filter oil out of the air? Right, and if you're trying to filter oil out of the air, you actually have to have a more powerful fan. So that's one of the big trade-offs is that the more powerful the fan, the less battery life you have. So the battery life on this is two to three hours per charge, which is why you have to have the hot spot to fit in a hot spot. Whereas the N100, the bigger fan actually, it's the amperage draw is about a fifth. So you can actually have it run for a good eight hours, or a good full shift, and you'll be fine. Using the same logic, the same everything. Yeah, so this is N100, this would be if you're just, like if you're in a forest fire situation, thank you. If you're in a forest, if you're trying to like, you know, deal with your periodic forest fire, air pollution, whatever, you can just use this. But if for whatever reason you're near a chemical spill, or like an oil spill, or something like that, then you could use something like this. And so, no, so on this, this is using the first germ guardian filter, and you can also see like that we have the different attachment process, that's another thing that we've iterated on, is attachment also things. And so, this one is N95, not N100, because it's using that HEPA filter in the box. The other, sorry, just real quick, I would say the other thing that we've been looking at is other masks, other than 3M. This is a mask from an Australian company called Active Mask. And we've been working with them, we had an interesting meeting with them earlier this week where we were like, your mask is not quite up to snuff. And they're like, yeah, we know, we know it's not. Like, because they were like, you should distribute this mask in the US. And we're like, well, it wouldn't pass NIOSH standards at all. And they're like, yes, we, now we figured that out. So they're going back to the drawing board, but because what we want to do to, like we need to get something that isn't a 3M mask, right? So we want to, and these guys are very positive. Are they open source? Not even kind of, right? But at least they're willing and amenable to working with us to produce something that we can attach to. We have a distinct sense that if we were to talk to 3M, they'd be like, what are you doing? What's going on? Like, no, don't do that. So maybe they would be fine, but we just haven't told them. Yeah, but going back to the level of filtration. So if you see, if you saw our diagram or basically coupling a fan to a filter, right? And so it's just the quality of the filter. So one of the nice things about our model is that you can make a custom filter input on top of the model. Yeah, and you can add whatever filter you can find. So the thing is, is that it's a matter of finding the right filters and the materials for your particular application. One of the issues that we came up with was the fact that we, I think I mentioned that we were trying to go for non rationed materials for COVID. And that's why we went with the HEPA filter as opposed to using, you know, like actual cartridge filter or something from 3M is because those were unavailable. Yeah, you couldn't get them, right? And there are PAPR units out there. There's actually a company that, there's a PAPR unit produced in Canada that will take the 3M cartridges and tie to it. Well, the guy that was, you know, you can order one from him and it's, I think it's open source, though. Yeah, it's not regulated or it's not been approved for anything. But again, it's the filtering material. We're not making filtering material and the level of filtration just comes from, you know, what filter material you're gonna use. Yeah. We have not done that. What we're trying to do is make, we're trying to balance to see if we get enough. Right now, frankly, we're trying to find the right fan so that we can balance the amount of power with just getting enough flow throughout. So we're not measuring the drop across the filter when it gets to a clogged state yet. Yeah, we're looking at sort of active sensing also, but that basically ramps up the cost of everything if we put in a pressure sensor for the flow. For the right pressure sensor? Yeah, I don't need 3D printed materials. If didn't you try to manufacture those? Now the low material is around the 150, 200. Yeah, and the other thing is that we made this so that it'd be printed on FDM printers or FFF printers, right? So the design choices that we made for it to be printable is different than what you would want if you were doing an injection molded or some other more commercial housing. And so that's the interesting thing, it's like you also have to take into account the nuances of the printing technology in the development of the plastic pieces, like orientation of print and the thickness of walls in terms of what you can get out of your nozzle, stuff like that. And we've done a lot of iteration on that to try to get a consistent print that works. Right, no, no, that makes sense, that makes sense. It's a good idea, because that will test for leakage. The question is, I mean, there's gonna be some inherent leak, the question is, is the fan gonna over, is the fan ever going to, is the filter ever gonna be included enough that it would actually activate the actual leak portion of that? Yeah, well we could just do that test. Right, right, that's a great idea. No, that's great. Yeah, no, no, that's excellent, I like that. We'll put that on the list. Yeah, yeah, yeah, yeah. Or design ideas, we'll take both. The question is, is the battery low? Is that at a percentage? Yeah, so I measured the battery discharge curve and then we decided that at this voltage, it's at 90% depleted and the buzzer goes off. Pizzo electric, Pizzo? It's a little Pizzo electric buzzer. So you're not doing coulomb counting, you're actually using the voltage curve, which is really, really tough on lithium batteries because it's a darn flat, right? It falls off the edge like poof. Yeah, we're just duplicating what Milwaukee is doing on their end, on the battery meter. Design change right there, I think that was the comment. Cool, no, I mean, like this was great. The discussion is what we're looking for. We've mostly been constrained by eyes on the project, right? We would love to have more people involved, especially if you can help contribute on any of the development side. Most of us, we've been doing this on the side when we have free time to do it. Or in my case, when I'm sleeping and printing in my dreams and also in real life. Yeah, my friends play video games, I play Fusion 360. It's really getting to the point where it's getting kind of obsessive. So looking for help. But please contact us at tetraby.io or check us out on GitHub also. And thank you. Any other questions or comments or offers of help? Money, right? Most of us are LA-based, but Patrick's house is actually. Yeah, yeah, yeah. We will be posting his address at the end of the day. Just show up whenever you want. We stream the bill parties and then we have to come up with some mechanism of testing to make sure that it's live. And so we'll try to do some of that. Like in our last stream, we hooked it up to the bench power supply to just show like what's the amperage like. Everything's working electronically, stuff like that. And then I did some just dance with the thing on because if you do it that way, then you can show that you can actually do physical activity while wearing the thing and not pass out. So that's a, there's probably more scientific ways to do this, but we're also streaming. And so I got to appeal to some kind of audience. And so this way my kids can watch and say, daddy, do some black pink and we'll keep it moving that way. So anybody want to take bets on whether the kids are actually saying that or if they're like, dad, come on, don't send this. Just kidding. He actually does really well. I mean, like I can't, he was trying to get me to do it and I'm just notoriously bad at those type of dancing games. I'm having flashbacks of my friends playing DDR and me standing there going, what the hell are you doing? Yeah. No, I'm a swing and salsa dancer. So it's just like, I can't do it. Yeah, so any other questions? There were a lot of interesting things in the pandemic that happened that were not like that I was totally unaware of and then I talked to Dana and was like, oh shit, is that really what's happening? Good times. So yeah, hopefully we'll see what he's got to say in half an hour. So thank you all very much for attending the talk. I got you now, sorry, my volume was turned down. I'll start talking. That's the hand mic, right? Mic check, mic check, mic check, mic check, mic check, mic check. Check, check, check, check, check, one, two, one, two, mic check, mic check, one, two, mic check, mic check, one, two, mic check, mic check, mic check, mic check, mic check, mic check, mic check, one, two, one, two, checking the mic, mic check, one, two, one, two, one, two, mic check, mic check, one, two, one, two, checking the mic, one, two, one, two, mic check, mic check, mic check, mic check, one, two, one, two, checking the mic, mic check, one, two, one, two, mic check, Mic check 1 2? Mic check mic check. Does it sound all right? Is that good? Higher lower? Higher? Higher. Yeah, mic check mic, check. How does that sound? Is that good? No. OK. Mic check. Mic check. Mic check 1 2 1 2, testing mic check mic check 1 2 1 2, mic check mic check 1 21 2, mic check mic check 1 2 1 2, mic check mic check testing 1 2 1 Hello hello and welcome to the last talk in the open medical track here at scale 19x. Up here we have Daniel Stemman who is got a lot of letters after his name. I met Daniel during the course of the COVID-19 pandemic. We were both on a task force for the mayor of Los Angeles, a task force that met twice and then nothing happened from it except that he and I met. And from that collaboration he pointed out we had a bunch of very eager volunteers who all made a bunch of really bad assumptions about what it is that we needed to be doing in order to be helpful in the pandemic. And he was really helpful in pointing out all of the bad assumptions that we were making. And this talk is one that I have beg broad conjoaled him to do about one of the problems that we saw at the beginning of the pandemic or that Daniel saw at the beginning of the pandemic involving N95. So having said that, I will pass off to you, sir. Excellent. Thank you, sir. Hello everyone. My name is Daniel and I'm a respiratory therapist. I manage a couple of teams at USC at the CAC medical center. And so the pandemic was very exciting for me. On March 16th at the beginning of 2020, that is when we had our, I believe, formal shutdown of all non-essential businesses. And so my wife's office in Pasadena closed. And we had a talk about what to do. And so we decided that she, who at the time was pregnant with my son and my, I had a three and a half four-ish-year-old daughter. Oh, you'll need those. Decided to move to her parent's house in Camarillo and I decided to move into the hospital. So this is just a recent photo of us. Very interesting couple of months after that. So, of course, you all know, you know, the significant surge of infections and huge demand on hospitals across the nation. There was a big delay between how bad the volume of patients entering the hospital was from the east coast to the west coast or like starting Washington and New York and didn't get to LA really till more like the summertime, closer to the beginning of July is when we were really very burdened with COVID patients. So, Niosch and CDC, who worked very closely together, decided that there was definitely going to be a really big problem and a shortage with a lot of things. But one of the main things that was of concern was protective equipment for the people that were going to be taking care of the patients. So they pretty much overnighted a very, very long list of masks and already registered respirators within the system that were international and decided that it was worth just kind of automatically making a bigger list so that we could have access to devices that were approved. That would make sense because they wanted to avoid a big kind of crunch panic for the people entering the supply chain at all the health systems, which kind of ended up happening anyway, but there was an effort to try and be prepared for that. So when that happened, many, many, respirators entered the system for healthcare that probably had not ever been used there before. We really didn't use N95 even for like my team being like the respiratory therapists or dealing with maybe the patients who have an airborne disease most frequently, really that often. So as an example, the academic medical center who would probably use PAPRs or N95s maybe 10 times a year, so very infrequently. And now we were going to use with all these new devices and so whenever a new device enters the health system, the regulations from the government through OSHA say you have to fit test the people to make sure that they fit them properly before you allow them to use it. This is one of the regulations that was waived when there were no masks available and there was no real feasible operational strategy to deal with fit testing prior to implementing. So it was waived for a short period of time in 2020, but the normal procedure would be to fit test anyone who would need to wear a mask. So I'm sure you're thinking like, oh, that makes a lot of sense. You don't just give somebody something that's supposed to protect them without making sure that it does, especially if you could die when you're doing your work. All right, so I have a very short regulatory joke because I have a graduate regulatory education, which means my wife thinks I am very boring. So the funniest part about the slide is that it is from 2001 and now that number is 1.8 billion. Thank you for enduring that. So my health system is about five hospitals and we have at the main hospital 84 ICU beds. We have 5,000-ish employees and at this kind of crunch time where all these new respirators were coming to the health system, we had to find ways, creative ways to fit test people kind of rapidly over and over and over again. And so I kind of volunteered to help out because the employee health team was so small. In fact, they were less than 10 people when the pandemic started and by July they were up to like 40-50-ish. So there's a huge burden just to try and make sure we were using safe equipment with like our nurses, doctors, respiratory therapists. We also really limited the number of people that could take care of COVID patients. So about 2,000 RNs are involved, 160 of my staff and overall though with all the physicians, trainees, students, it's about 5,000. And probably the absolute worst part of our burden was that for PAPRs, which are recommended for any aerosolizing procedure, which means anything in respiratory therapy basically, any aerosolizing procedure you have to use a PAPR, that is the recommendation, because of the risk of exposure in that environment is a lot higher. So if someone is having a procedure in the airway and we're jetting air or medicine or cutting or using a bovian surgery or really anything, it aerosolizes particles at a high volume and so you have a much higher risk of getting sick. So at the beginning, we had nine for the entire campus, which is really not at all enough to sort of manage procedures in the ICU and definitely not in the OR. And so we had to come up with a lot of creative solutions. And one of the things that we did was we made a cart that had four PAPRs on it. And we had just a team of people that would run it all day, every day, everywhere, all the time, it would just kind of run around procedure to procedure. And then we identified particular providers who were going to be in charge of intubation, since it's a really highly aerosolizing procedure and a high risk to get sick. That person would wear it all day and would try to do all of the intubations in the OR, which had to change like all of our workflow and a lot of different things. Okay, talking a little bit about the problem with fit testing. So as Daryl alluded to earlier today, the most common method used for fit testing employees in hospitals and most places, with the exception, of course, of BSL3 level and higher labs, they use qualitative fit testing. So we're using the joke from earlier. If the test is looking at whether or not you can taste particles and we have literally millions of people infected with the disease that causes you to lose smell and taste as like one of the more common symptoms, it's kind of a no-brainer that this is going to be a problem. So I was really concerned about my team and about the people that I was working with, so much so that I went and stole a bunch of equipment from the university, which was fun. And we rated the BSL3 lab that was shut down for a little while and took a bunch of their stuff. And we found a port account in the employee health center on the main campus, brought it to the hospital, and we kind of like started up our little corner of investigating student health. Yeah, sorry about that. Yes, sir. A respirator is a passive mask. So NIOSH has some interesting language. They refer to them as filter media, filters, respirators, N95 product. But respirator in the sense of what we're talking about in this talk today, we're talking about the passive N95 mask. Yeah, please. This is interrupting, by the way. Go ahead. They do. Correct. So you would think that. But actually, there's a lot of data out there and papers on the problems with qualitative fit. One of them being that people just don't take it that seriously. And they don't really understand it half of the time. Most of them when queried after an exercise thinks that they were supposed to smell something. So there's actually a lot of just a gap in general with people taking it seriously. There's also a huge gap in its efficacy. I think it's pretty well documented that it's only about 60% effective. So that's not very great when you're thinking about the pandemic. But yeah, thank you for bringing that up. It definitely has more than one problem. In our observation at the beginning when we decided to start looking into it ourselves, we teamed up with a few people at different universities between Caltech, USC, and UCLA. And we made a very long list of the masks that we thought would work well and then those that would not. And we found 173 different approved restorators that didn't seem to work so great. So what are we supposed to do about this? Yes, sir. It's a really good distinction. Yes. So I am talking about it from the lens of the filter media all working perfectly fine. And by working perfectly fine, I mean the filter material itself would capture 100% of 40 to 70 nanometer particles all of the time. And so the filter is fine. And the problem is that everyone's face is really kind of like a fingerprint. And trying to have sort of these simple standard shapes really don't necessarily fit as wide of face shape variants that there are just out there in the world. And organizations like NIOSH actually have regional mannequin face types. And so they have seven regions of the world with an average face shape and then five sizes of those average face shapes across the seven regions. So they do their best to try and make it easy. So July 7th, we received a word that we were going to be getting multiple pallets like something like 20 million units of this particular mask from Make Right called the 9500. And so we had established by this time sort of an internal protocol of using quantitative fit testing as a gatekeeper for implementing new products. And we found that we couldn't make it work really for anyone, meaning it did not fit anyone's face that we put it on. And if you're familiar with this particular mask, it would be like if you saw it a globe in half and it was almost like a perfect circle cut in half with tabs on it. So this shape is really rigid and really not not really forming well to the face at all and just always on almost every person would have huge huge open gaps, which for the people that are going to go into the ICU or into the clinical environment was like obviously not a good thing. My first assumption was that this was a manufacturing problem or we had received counterfeit product or we had some problem that I didn't have a good way to measure. And so looking at how we would send the message out to the proper authorities or get help from the right people really just landed on using MedWatch as a med device reporting system as really the go to since that's what we do most often in the hospital. But in reality, that probably was the slowest approach that we could have taken to the problem. But that's what we knew how to do. Within a week though, I was able to catch up with the local regulatory board in California, which was called Cal OSHA. And we made actually this slide deck, but shorter to explain our situation to them. And at that time, they told Mark and I, yes, we know. Thank you for working on this. Could you please call our friend? And this was the deputy chief at NIOSH. So we said, sure, absolutely. About a month later, we were able to meet with Colleen and explain our situation to her and her team. And she also emphasized with this situation as not new news, but something that they had been working on since 2008. And unfortunately back then they had lost a lawsuit with the lobbyist around changing the rules at NIOSH to implement new requirements for manufacturers because it was going to be very expensive for the manufacturers. And there was really at the time probably a good argument that there wasn't really a good burden that made sense that they needed to spend all this extra money and make these big changes. But high insights 2020. So there were some questions earlier about the regulatory agencies. So I made this slide while Darrell was talking. So the CDC, they make a lot of recommendations to the public and they work alongside industry and a lot of professional organizations to kind get information out. NIOSH actually tests and validates that a industry or a sponsor who's submitting a product not only is meeting all of the GMP and the other sort of industry requirements, but they are actually sending to the agency what they're going to be making and sending to the public. And so they have procedures for validating. Unfortunately still to this day there actually is not a strict validation for FIT. It is a sort of gray area of you assert that you have implemented one of the protocols either qualitative or quantitative. They have a procedure for doing those protocols that you perform and then you submit. But NIOSH does not retest and validate that your FIT exercise is, you know, true to realize there's no validation testing for that. Okay. FDA, they do a lot of things. But essentially they're really built on just two words and that is misbranded or adulterated. Everything, you know, from the FDNC Act is based on those two words. So they do that across a broad swath of things from med devices to drugs to biologics to food and cosmetics and other things. Okay. So the masks don't fit. So what are we going to do? We're just not going to use them or do our best to not use them. And we, through this internal process of testing and eliminating options, really, really, really narrowed down the sort of masks that we were looking to buy, which was really stressful because at that time we had already purchased some masks that were not going to work and so we had to figure out what to do with those and how to get them, you know, maybe get our money back for them or figure something out or sell them or, you know, this comes to like an ethical problem because if you're selling something you decided not to use, you're like... So after about 18 months we were able to return them to the manufacturer. I think one of the more depressing moments when, you know, reliving that timeline is we were sure that this was a very large problem and that we needed to try and help as many people as possible with this problem, but it was a huge operational lift and a really, really big like believe it issue as well. And so while we did our best to try and like tell our friends, and I mean friends, I'm talking about like other AMC's and systems, you know, most of them just it wasn't feasible to sort of implement the protocol that we had decided to implement. We're sort of a kind of living in a bubble as well because my main hospital does not have an emergency room so it can't really get flooded with COVID patients like a normal let's say county hospital. We have a county hospital which was a different situation. So yeah, at the end of August, which is sort of the tip down on the first peak here in LA, you know, our infection rate for the providers and staff that were working with COVID patients was only 0.27%, which is insanely low. That's crazy low. That's lower than probably everywhere in the entire country. And if a lot of places are not measuring it, so it would be hard to like prove that, but it's very low for those we compared ourselves to locally. And places like Cedars and UCLA and UCSD or UCSF or Stanford or other very large systems were all above five and some of them closer to eight. So many, many times the infection rate that we were seeing and at that time we felt like now we need to like make a bigger deal of this. Yes, sir. No, no, please keep doing that. That is accurate. Well, that is not accurate. They do not have a requirement where it is written into the rule that they must validate the fit procedure that is submitted to them. But they do have the ability to test it with ASTM now. Correct. Yes, it is a lobbyist issue. So it's a money issue. When you want to change rules in the government and it creates a huge financial burden for manufacturers who need approval from the government to sell their product, you have to make sure you have a really good reason for that because they will throw a lot of money at making sure that either that is basically accurate. Yeah, and that would be around like the Ebola timeline. Yeah. And I don't want to say that it's like so direct like that. It's a little bit different, of course. Concerns were raised by those at NIOSH about their rules lacking a strict requirement for fit. And so they wanted to, you know, come to consensus with industry around what would make sense and how they could work together to get to a place where there is a strict requirement. And they thought they had that. And so they submitted that, went to the federal register, went to public comment, and then it got sued and lost and never was implemented because they lost in court for the game against the cost was worth the burden. Although now, you know, we have the hindsight of 2020 thing. So limited options. We were looking at, as I said, 173 no longer options for us and had to try and stick to the very few products that we knew fit best in our workplace. That's probably the most accurate way to say that. And so we were looking at a short list of about five different respirators made by 3M and Honeywell and Halyard. And the 3M respirator that we chose to buy in bulk was actually very contentious at the time because it has a valve on it. But we decided that it was worth prioritizing the safety of the staff over the off chance that we had people who were infectious who didn't know and were potentially passing, you know, contagious particles into the environment where everyone already had COVID. So it didn't make too much sense to be worried about it. And so we decided it's better. This fits better. It's safer for our people. Let's do that. And we just kept it in the areas where, you know, all the patients already have COVID. And we didn't use them, let's say in the OR, where, you know, we already have a concern around, you know, people wear surgical masks in the OR during surgery because we don't want them spitting into people. So you wouldn't want a mask that had a valve on it because you could, you know, accidentally get, you know, oral secretions into, you know, maybe a surgical site. All right. Math to the rescue. We now need to figure out, because we failed at, you know, changing the rules, because that's not an overnighter. And apparently there's a long history of problems there. We needed to figure something out. We needed to do something, you know, we knew too many people who are having problems, who don't have access to anything else. And so we're like, okay, how can we make use of what they do have access to? And so we decided to collect our, you know, group of hoodlums who are good at the math and the beep boops. And that would be some of the folks here today and some of the folks who are maybe online but not here today. I needed major beep boop help, so that is not my thing. Available to us were some very smart people that were working on solutions for other things like oral implants for dental implants. And there's a company called Bellis 3D who had come out with an app that would provide a file that would create a fingerprint-like frame for your face that would increase sort of like finger pressure around a surgical mask. And the idea behind their product was to reduce community spread by making regular surgical masks fit on your face better so that like the leakage wasn't as big of a problem. And so we immediately thought about like reverse engineering this idea and decided to start, you know, we contacted them, we spoke to them a few times, we got some software setup and I got one of their little cameras with the little, you know, beep boop thingy on it that like, you know, checks stuff on your face and creates the file for you. And so the problem with that though is that since everyone's face, as we talked about earlier, is kind of like a fingerprint, it's really an operational nightmare. I mean you have 5,000 employees that's like 5,000 faces that have to be scanned and then there needs to be some bank of printed frames that would like be available when they broke it or whatever. It wasn't a really safe idea or an efficient idea, but it did work decently with the exception that their frame, since it was trying to lock air in, actually had a major leak problem down below and so it wasn't perfect anyway. So we kind of needed to start from scratch with something that was going to work across as many faces as possible as quickly as possible. So here are the team members involved from the beginning, sort of the same group of people working on a few other projects at the same time. And this is a rendering of one of those frames from the early days that we ended up using for the paper that we published this in. And what we did was we took the NIOSH head forms and made estimates for what would be sort of the closest to the middle across a couple of the head forms. And to come up with a shape that we thought was a good starting place to start testing. And so this is an image of what that looked like. Oh no, there was a delay click. Sorry about that. Here's a couple of images of the early frame. One of them on the far left is just the bellis frame on my face. Those in the middle are our first version of the new frame that we created. And it might be hard to tell from the images, but if you look in the lower section on my face there, you'll see it's kind of like a lot more hard angle pointed sharper angles. Whereas the one we created ended up being much more rounded and almost flat and longer along the sides, which would go further along the face and up the cheek. The frame that we ended up using for the testing is there on the right. And that's on one of the Honeywell masks, the 300C or something like that, and the DE2322 BYD Care Mask, which was one of the largest volume masks distributed out of the Port of Los Angeles to healthcare workers. This is some of the data from our paper around what we were seeing in the days when we knew we needed to start working on a plan to fix the problem. And so we have the 1860, which is the surgical 3M mask. That mask fits incredibly well. It already has multiple sizes, and it's okay to use in the surgical environment. So that's why we already were using it in the hospital. That one turns out fits really great. Across all of the quantitative testing protocols, it was really awesome, like best scoring. And then you'll see there in the closer to the middle, you have the 8511. That's another 3M product. That's the one with the valve. That one also averaged passing scores across all the eight exercises from the OSHA requirement, the protocol for fit testing. And so you see why we chose that as the bulk item to stock the health system with. In the middle, the 9500. That's the one that I received in July. Yes, sir. So these numbers represent an aggregate score from the software of the TSI Port Account system. And it is called a fit factor. It's sort of a moving target based on particle leak. So higher is better. Yes. And above 100 is passing. It's between zero and 200. Yeah, it's actually between zero and 5,000. But 200 plus is where when you put it in in 95 companion mode, what it does is it changes the voltage on the optical particle sensors so that it focuses on counting only 40 to 70 nanometer particles. And when it does that, it's assuming that if the mask fits you that there will be zero particles of that size sampled from inside the mask. So the way that the system works, it's got two sampling lines. One of them is connected to the mask via a port that you create. You punch a hole. The other is sampling ambient air. You saturate the room with a minimum of 200 particles per cubic centimeter. And once you're at that saturation, you can then proceed with the testing. And what it wants to see is that none of those 200 or so particles per cubic centimeter enter the mask at all throughout the testing protocol. Because if they do, that means 40 to 70 nanometer particles somehow got through the seal on your face. Does that make sense? So the DC 300, which is the Honeywell mask, it did work a little bit better. But this is the mask that you might remember was made in Arizona at that factory they stood up for COVID. And like the president went there and they were like talking about it. But that mask shipped with no foam seal and no wire, which really kind of doesn't make a lot of sense. Because if you're going to use that in a hospital, every time you move, it's going to leak. So it really struggled during the exercises where anyone had to do anything. And if you know something about the fit protocol, you have to turn your head side to side, you have to move your head up and down, you have to talk, you have to bend over. And anytime you did any of that, it would automatically start to fail, even for the rare face that it did fit on. And it really seemed like it was made for like a Viking human that was like seven feet tall and large, because the mask really only fit really large people. And then on the last column there, where you can see everything is averaging into the green again. That is the same mask, that BYD care mask, the pointed one, that a lot of people still wear and are mimicked after the KN95. That is our frame on that style mask. So you see it went from failing miserably, or about half of the expected protection to well over the passing range. So we thought, yes, excellent, great, overnighter, yes, we've just solved all the healthcare workers getting sick from the pandemic. That's not actually what happened. All right, here's just some other illustrations of in this table, you know, the passing line is right there in the middle 100. You were asking about what the numbers meant earlier. So the three columns below are, you know, labeled probably very small, difficult to read, but the first one is the 9500. That's the Make Right Mask that I received July 7. The DC300 is the red one. And then the DE2322 without the frame is the purple one. So the three above, the yellow, the light purple, and the orange are the two 3M products we talked about. And then the folding style in 95 mask with our frame on it at the end in orange. So it was just about performing as like the normal masks were performing. So who thinks we were able to just give it to everyone and use it in the hospital? What might be wrong with that? There are some rules about emergency status. So if you start to use off label or unapproved products in your health system, you have to declare emergency status, meaning you cannot do anything like normal. You now have to constrain your entire system to operate within emergency only, meaning you can't do most surgeries. You can't see most patients. You can't really do anything normal. You can't even see cancer patients. There's a whole lot of things that you are now do because they don't want you doing things with equipment that isn't validated and that makes perfect sense. But from a feasibility real life standpoint during the pandemic it didn't make much sense. So if we were to implement them and just shoot them across the nation for people to use with those folding masks, that would have basically put any hospital that used them out of business because they wouldn't have been able to pay their employees because they wouldn't have been able to see any patients. And then it kind of doesn't make any sense. You kind of have to stay open and see patients. Okay. So we met with Calishe again. We shared the information. We let them know that we felt like we had a solution and we wanted to know what the quickest regulatory pathway to approval would be. They quickly said we don't do that. And please call Nyash. So we did that. We called our friends. We've been talking to you for about five or six months at that point. Meeting with them monthly to kind of go over how things were going. And we shared with them, hey, we have a solution for this one style product in this category where there are tens of millions of them everywhere. All police, all fire, all EMS services in California. Most of the county hospitals, the VA everywhere is using this product because it was bought in the tens of millions by the government. And so they let us know that because it sort of meets the definition for respirator accessory, it can't be used with any mask that it's not submitted with. So what we needed to do and what we tried to do when we called the CEO of BYD was to request that they resubmit an amendment to their application with Nyash, which was emergency authorized without any data, to include an optional accessory, which is basically like the most benign way to say, hey, your thingy doesn't work. We have a thingy that helps it work better. Please don't say that, but allow us to add it as an option and then people will have thingies that work better. And so that ended up not happening, unfortunately, but we tried. And Nyash really tried to help us out to partner us with manufacturers that might be open to that, especially those who were working on creating local manufacturing for the future because that's still ongoing. A lot of companies are trying to create sort of within the continent supply chain, and that makes a lot of sense. So they were partnering us up with those people and we still work with manufacturers who are trying to do that and stand them up. We ended up publishing it, so some people may have read that. And we're still working on this. And the good news is that in November of 2021, Nyash resubmitted their proposal for the rule change, which includes this strict requirement for quantitative fit. There, of course, will be some looming battle in court at some point, but for now, in terms of public comment and otherwise public engagement, there's a lot of support for making a change because it makes a lot more sense today, probably than it did in 2008 when it failed. They also released new guidance and in their new guidance documents, they referenced the proposed rule and they talk about how manufacturers should prepare for future requirements, including strict quantitative fit testing. And that is the end of my presentation. Are there any questions? I will add one addendum to Daniel's presentation because he and I have been trying to talk about this and I failed in this one. Part of the reason why I think this talk belongs at scale is the fitra frame that he was describing is available. The STLs for it, if you want step files, we'll get you the step files. If you want to print your own or if you want to start iterating on it and coming up with your own versions of this, they are available on github.com slash tetra biodistributed. And if you need me to write that down for you or whatever, I can text it or give it to you. So, and I would... I need you to know me and then I'll forward it to you and you can forward it to them. That'll also work. We do also have the data that was used in the fit testing stuff, but we are not allowed to give that out because there's people's names in it. So, we will not... We have the data, but we cannot open source that. So, I hope you will bear with us on that one because there are laws, let's say we cannot. So, having said that, sir. Yes, sir. Qualitative versus quantitative fit test? Sure. You know, I got a qualitative fit test kit. Okay. Because you can buy them right for small business. Amazon. Yep. 3M fit. 3M FT10, I think is what I got. Right? And the 3M one is way better. It's got a way better hood. Sure does. I had an Allegro one before that and it's just a bag you put over your head. The 3M one is the one we use in the hospital. Yeah. So, one of the... You know, I mean if... And it's a pretty simple procedure. So, I'm trying to get all my friends to do that. It's trying to get a professional come by work. But, and I've read about the quantitative fit test and I'd rather it's better. But, I mean it's a lot more expensive, right? I mean the fit test kit I got is a couple hundred bucks, right? It's pretty cheap. Right. The port accounts are like 10 grand. Yeah, they're more. The port accounts are... Depending on how many you're buying, of course, they're about $14,000 to $15,000. Yeah. With like a PM package and service and all of that. Ah, okay. Well, I mean somebody was offering to sell me one for 10 grand. Yeah, you can probably... Probably without the support. You can probably eBay the older ones for maybe a thousand, actually. Oh, if you want the ones that aren't calibrated, those are way less than a thousand. You can pay to have them calibrated and you can mail it. Yeah, you can mail it to Minnesota where they do all of the calibrations for TSI, like globally. Oh, really? Yeah. In fact, I recently was working with one of the TSI engineers to calibrate a 8130A, which is the device for filter media testing, for penetration testing and other testing. Yeah, they actually have really stood up a lot more in terms of service for the legacy products to be calibrated and make sure they work since the beginning of the pandemic. Nice. Yeah, yeah. One thing I work and I'm trying to get a group of friends together to get some professionals out to do this just because it's like, I'm a guy. I have the kit. I read the instructions, which actually, I think, is all that OSHA requires. Not everybody's sure, but I think so. Yeah, so it's the OSHA CFR 291910.134. That section got both fit test protocols in it, and it outlined very in-depth each step. You can probably get it on a .gov website. But you could also go to NIOSH's website and download their most recent guidance, which is from June 29th. And it also has those sections in there. They have SAPs for everything. Okay. Yeah, I was just following the instructions from the kit because I figured 3M probably wouldn't steer me wrong. Definitely. The kit comes with proper instructions. Yeah. All right, cool. But what you're saying is that there's a big jump up to the quantitative, right? You're saying that's way better for... There's papers that say efficacy is as wide as a gap of 40% so that you could have as much as a 40% error rate using qualitative. Because taste is a difficult metric. Yeah. Yeah. Even with the whole idea of it being very sweet or very bitter. Yeah, yeah. And they are, oh boy. Yes, they are. But it is still tough, especially when the procedure starts with seeing how much of a dose makes sense for the person you're about to test. Yeah. They have now just tasted it. Yeah. And now they're so sure what it tastes like. And you'll notice that after that point, they're no longer sure whether or not they're tasting it. Yeah. And there's this sort of subjective game of, I don't think I taste it. I'm not sure. Yeah. Yeah. Yeah, a bunch of my friends, we did it like two or three times before we were like, oh yeah, okay. Yeah. But... Okay. Yeah, yeah. Anybody else? Any other questions? So one of the other problems with the qualitative taste test is that it actually doesn't challenge whether or not the mask material itself is N95. So one of the things that we discovered while we were doing this testing is that that qualitative test just tests for leak. So if the mask is leaking, what happens if you're wearing a counterfeit mask? You wouldn't know with just a qualitative test. And with the quantitative test, we found a... More than 30 times we had counterfeit masks sold to us by big distributors. McKesson, Medline, yeah. Well, keep in mind, this was during the shortage. I mean, they're going through their normal checks and balances. It is really good counterfeit products. But just keep in mind that that test doesn't test for the mask itself. And that's actually a very important distinction because we found online there was a lot of misinformation about that because yeah, there was a huge amount of misinformation about that. People were suggesting have a receiving validation where they have a quick tester. Who does? Like the Port of Los Angeles as an example. So they would open a random box and then put a mask on it and then run a quick test, quick penetration test. And so then they would have some data. That's usually like the top one. If the top box was normal and then the rest are not, you know, it's kind of not impossible to sort of fool. Yes, ma'am. Yeah. So perhaps I missed it in between. But from what you said, there seems to be like two extremes. There's one kind of mask that fits a seven and a half foot tall viking. And then there's a set of masks that fit just sort of like, you know, of rain. Oh, and then there's like the perfect fit that is like literally individualized to an individual face. What is a good in between that still fits a range of faces because frankly, looking like I do when I hear, well, we have like regional fits that sort of fit the average face for that region. I'm like, well, that's never going to be me. Right. Yeah. So that's why OSHA has the requirements that they have that you at least in the working environment, which is the only environment they have control over that you have to test that it is a good fit for your employee. So that's why there's so many because the many options really do cover a lot of different faces because they are kind of, you know, quite a bit of them are very different. So together, you know, if you had a lot of them and you could select the best one for each employee, that's really how the system is supposed to work. That's sort of not totally reasonable. If you have to test, you know, 173 of them, like we did something like 240 different respirators through our little bank of testing. And, you know, no system anywhere that I know I was going to have that many different products, they're going to have maybe three. So what ends up happening in the hospital where you're concerned about, you know, them definitely getting sick, you would either not have them work in that environment or have them wear a higher level of protection, like a papper or another device or an elastomeric mask, or, you know, you find a mask that fits them. My question is that like 3M, they have small, medium, and large. They do. They work with, actually 3M works with the same engineer that designed the different sizes for tampons because from like a biological calculation experience, there's not that many different devices where you had to use masks to figure out how many different sizes made sense for a particular, you know, fit. And so what they did is they have three here, and other parts of the world, they have five. And in other parts of the world, they have two. So they appropriate the like amount that makes sense for where they are. I don't know if they're all still like that, but most of them are three. But it isn't just the one like, you know, fits one. They have like tiers and they follow the regional map as well. Because NIOSH has small, medium, large, extra wide, large, and extra narrow, large. Right. Yeah. Which is essentially tall, the extra narrow, large. Right. Well, yeah, the, the DC 300, the one that we tested a lot, that one is very wide and very tall without a wire and without a sponge. And so it really only fit very large faces. Any other questions? We did. The Honeywell Mask? We did get it to fit a couple of people. They didn't, not all of them made it into the dataset, but I think we fit like three people out of the probably more than a hundred that we tested. Yes, sir. They do not have a rule about that. They do expect you in your quality system to explain why you have the pressure that you have, the length that you have, and the material that you're using. But there is not a specific rule about how much pressure should be or need be applied. There's a range of materials you ought to think about. Elastomerics are being made out of different plastics. Some of them use silicone. Some of them use gel. And then a lot of them are just using like polypropylene folds and cotton. So all of these different materials are being validated by the same team. So it would be hard to make like a pressure rule apply to all of them and it makes sense for all of them. Most N95s do not come with adjustable straps. Most of them do not. Some of them do. A lot more of them do now, especially KN95s that have the external straps where the strap attachment is not in the breast zone. It's like on the edge of the mask and it's either glued on or sonically welded. Those ones, a lot of them will have like the pull strap like you see on the Elastomerics. But that's, I see that, or my understanding is that is more of a new trend than, you know, before the pandemic, I've never seen a N95 at least in the hospital or in the OR where it was adjustable. The Elastomerics are adjustable. The Pappers are adjustable to fit your head, but they don't have straps usually. Yes, sir. The KN95s and stuff like that. Yes. My experience testing them is I've never seen anybody pass on one. Right. Okay, that was my only question. Why we did this project? They generally have great performing filter media, but they don't fit. Right. So the idea is that that's why we open sourced the fitra. So you can't use the fitra in a hospital environment because of all the reasons that Daniel elucidated, but if this is a topic that you care about, you could take the fitra frame and a KN95, print your fresher frame and then attach it. I will say though that with the KN95, with ear loops, I have made a lot of them work by just adjusting the pressure, shortening the straps, or even using very, very small, like even like a rubber band or a plastic loop to add the ability to adjust makes a massive difference because they generally have really good filter media. Yes, sir. Yep. When you what? Oh, I know it's sealed and it feels comfortable and I can wear it. This is so many a week of time. I guess you mentioned that there's a time to take it off and replace another one. So I do have two or three of them that I most of the on label recommendations for N95 for wearing them for extended periods of time is four hours at least in the hospital. And that's really kind of an impossible task. You can't really wear it that long. It literally bruises up your whole face. But the last America have a way better softer fit. They have a much larger surface area usually contact. So they can seal under much lower pressure because there's a lot more surface area in contact with the face. And they generally have a material that's like really pliable like a silicone like or a plastic or a rubber. And so those are more comfortable. You can wear a much longer, but they come with cost to, you know, you often can't use them with a lot of different equipment, depending on what your workflow is. So let's say you were like a ortho surgeon is definitely not going to work for you. You're going to have the inability to look into your instruments. Or if you worked in like a small surgical environment, you definitely can't use a microscope. You can maybe not use a robot. You're not going to be able to use depending on the interface you're using, you may or may not be able to look into the thing you need to look in because then your your face is hitting the device. All right, well, let's give another round of applause for Daniel. And that I'll say that this is the end of the first open medical track at scale. And thank you all for showing up and I really appreciate it. And if you have any further questions, we'll be around. Thank you.