A review of open source ventilators for COVID-19 and future pandemics

Coronavirus Disease 2019 (COVID-19) threatens to overwhelm our medical infrastructure at the regional level causing spikes in mortality rates because of shortages of critical equipment, like ventilators. Fortunately, with the recent development and widespread deployment of small-scale manufacturing technologies like RepRap-class 3-D printers and open source microcontrollers, mass distributed manufacturing of ventilators has the potential to overcome medical supply shortages. In this study, after providing a background on ventilators, the academic literature is reviewed to find the existing and already openly-published, vetted designs for ventilators systems. These articles are analyzed to determine if the designs are open source both in spirit (license) as well as practical details (e.g. possessing accessible design source files, bill of materials, assembly instructions, wiring diagrams, firmware and software as well as operation and calibration instructions). Next, the existing Internet and gray literature are reviewed for open source ventilator projects and designs. The results of this review found that the tested and peer-reviewed systems lacked complete documentation and the open systems that were documented were either at the very early stages of design (sometimes without even a prototype) and were essentially only basically tested (if at all). With the considerably larger motivation of an ongoing pandemic, it is assumed these projects will garner greater attention and resources to make significant progress to reach a functional and easily-replicated system. There is a large amount of future work needed to move open source ventilators up to the level considered scientific-grade equipment, and even further work needed to reach medical-grade hardware. Future work is needed to achieve the potential of this approach by developing policies, updating regulations, and securing funding mechanisms for the development and testing of open source ventilators for both the current COVID19 pandemic as well as for future pandemics and for everyday use in low-resource settings.


Introduction
Coronavirus disease 2019 (COVID-19), caused by a novel coronavirus (SARS-CoV-2), is in part so dangerous because it threatens to overwhelm our medical infrastructure at the regional level, causing spikes in mortality rates [1][2][3][4] . Within the medical infrastructure, there are critical technologies that are generally available, but simply do not exist in a high enough density to handle the excessive volume of patients associated with pandemics 5 . Thus, people die unnecessarily throughout the world because of a combination of COVID-19 infections and the lack of access to some of these technologies 6 . Ventilators are an example of technologies that are currently in critical short supply 7,8 . Mechanical ventilators are essential for treating both influenza and COVID-19 patients in severe acute respiratory failure 9,10 . Past studies have shown that intensive care units (ICUs) will not have sufficient resources to treat all patients requiring ventilator support during a massive pandemic [11][12][13] , and ethically challenging triage 14,15 would need to be used to decrease mortality over first-come first-served basis for ventilator allocation among patients. Some work has shown promise for using a single ventilator to support multiple patients during a disaster surge [16][17][18] . In addition, it has already been shown that 3-D printed manifolds can assist with rapidly deploying this solution and there are open source designs 19 . This is not necessarily straightforward 20 . Although some countries, like the United States, have stockpiles of ventilators 21 , there is consensus that there is not enough supply for serious pandemics [22][23][24][25] and that rationing would be needed 26 . The current medical system relies exclusively on specialized, proprietary, massmanufactured ventilators from a small selection of suppliers. This supply model clearly fails when there is a sudden surge in demand for a relatively low-volume specialty product such as ventilators in a pandemic as analyzed here. The vast majority of medical equipment is heavily patented by a few specialty medical firms that sell small volumes because during 'normal' times, a medium-sized hospital only needs a handful. These firms have historically aggressively protected their intellectual monopolies 27,28 to the detriment of human lives. In addition, non-practicing entities continue to attempt to actively prevent medical treatments from being deployed, even during the current COVID-19 pandemic 29 . Putting aside the absurdity of patenting and then obstructing others from using obvious inventions in normal times [30][31][32] , in the wake of a pandemic where millions of lives are at stake, it is intuitively obvious that this type of greed is no longer acceptable.
Fortunately, with the recent development and widespread deployment of open source small-scale manufacturing technologies 33,34 , there is now another way -mass distributed manufacturing [35][36][37][38] . In this new model, designs are developed and then shared with open source licenses freely on the Internet so that others can simply download and replicate the design on their own equipment, even at the household scale 39 . There has been tremendous and ongoing success of open source scientific hardware proliferation 40-45 , where lower-cost and superior-functioning custom equipment as compared to proprietary scientific tools [46][47][48][49] . Based on such scientific hardware results, there appears to be a significant opportunity to apply open source design principles 50 and mass-scale collaborative distributed manufacturing technologies to make medical equipment 51-54 . In the current situation, this would at least partially overcome medical supply shortages 55-60 in general, and specifically for ventilators [61][62][63] .
Of these enabling technologies, the most advanced is the fused filament fabrication (FFF)-class of desktop 3-D printers that have spawned from the self-replicating rapid prototyper (RepRap) project [64][65][66] . With the distributed manufacturing model, designs are downloaded even in remote areas and are manufactured on demand as needed 67 from readily available (and possibly recycled 68-80 ) materials. These 3-D printers are, in general, not particularly fast when making products, but with tens of

Amendments from Version 1
The manuscript was updated in several ways:

Introduction
Based on recent news about the Italian patent infringement lawsuit this example was removed entirely. Existing peer reviewed literature Added detailed analysis of new study, which is the first fully documented ventilator in the peer-reviewed literature. Increased discussion about need for diversity of solutions needed in a pandemic. Updated grey literature review, added the Read review, many of the other major teams and projects, and listed the approaches. Stressed the need to share plans Future work Added a section on best practices for sharing a design Included points about the need for multi-disciplinary collaboration, the need to involve medical personnel, and to aim to publish in the medical literature. Discussed the use of sharing incomplete designs and status tags. Added a paragraph on the need and current status for ventilator testing.
Deleted earlier claim about easier replication in developing communities and replaced it with a call to streamline regulation while maintaining standards. Added discussion of need for transparent quality control, standards and qualifications. Added concept of expanding Good Samaritan laws. Qualified conclusions that are changing rapidly.
Added references throughout text and fixed minor typos. For more granular details of changes see individual responses to reviewers.
Any further responses from the reviewers can be found at the end of the article REVISED thousands of 3-D printers already strategically deployed all over the world 81 , they have the capacity to fabricate an incredibly diverse and large range of products (growing exponentially) 82 , which have already been shared with open source design licenses. Here, the potential will be analyzed for hardware that can be as-much-as-possible digitally manufactured using accessible low-cost fabrication tools like RepRap-class 3-D printers and then readily constructed from widely accessible materials and simple tools (e.g. DIY hardware store sourced along with Arduino-class microcontrollers). RepRap technology in particular is stressed because designs in that ecosystem are already frequently shared, enabling true distributed manufacturing by making use of manufacturing equipment near the point of use. There are, however, many other means of open hardwarebased digital distributed manufacturing approaches including CNC mills, laser cutters, engravers, and etchers and other digitally controlled fabrication tools. As pointed out by Mohammed 83 many of these tools would overcome limitations of 3-D printing (e.g. speed of replication for flat parts are more easily cut from stock with a CNC tool using subtractive manufacturing than 3-D printing based additive manufacturing).
In this study, after providing a background on ventilators, the academic literature will be reviewed to find the existing and already openly published vetted designs for ventilator systems. These articles will be analyzed to determine if the designs are open source both in spirit (license) 84 as well as required practical details. To be open source a ventilator project needs to include: 1) the design source files (e.g. computer aided design or CAD), which are needed to iterate on the design mechanically; 2) as well as production files (e.g. STL files which are used by 3-D printers to make mechanical components); 3) printed circuit board (PCB) layouts and other electronics design files to allow production as well as design evolution of the electronics; 4) bill of materials (BOM), which is needed to allow reviewers to evaluate the components employed as well as more easily find alternatives; 5) list of tools required, which are needed to determine if a device can be fabricated in a specific facility; 6) wiring diagrams, which are used to assemble the device with electronics; 7) firmware and software, which are needed to run the actual device; 8) instructions for the assembly, so makers can fabricate the device when the parts are made or acquired; 9) instructions for calibration as in many cases ventilator designs demand fine tuning to achieve adequate performance; 10) instructions for operation, so the end users can use and maintain the device.
Next, the existing Internet and gray literature will be reviewed for open source ventilator projects and designs. Lastly, as this is a rapidly evolving area, future work will be described to enable wide-spread mass distributed manufacturing of open source ventilators to fight against the current COVID19 pandemic as well as for future pandemics and to provide the devices to low-resource regions of the world that are underserved even in normal times.

Analysis of literature
Oxygen therapy coupled with mechanical ventilation is meant to support patients so that an adequate oxygen saturation (>88%) in arterial blood is maintained 85 . The mechanical repository cycle has four parts: 1) inspiration, where the exhalation valve of the ventilator is closed and the ventilator uses pressured air to cause gas to flow into the lungs; 2) cycling, where changeover from inspiration to expiration occurs; 3) expiration, where the main ventilatory flow is interrupted and the exhalation valve opened to allow gas to escape from the lungs, and 4) triggering, where the changeover from expiration to inspiration occurs. According to Andreoli et al. 85 , mechanical ventilators are classified on what factor terminates inspiratory flow, as follows: 1) pressure-cycled ventilators terminate flow when preset pressures are reached in airways; 2) volume-cycled ventilators provide a set volume of gas to the patient over a range of pressures (but a maximum pressure is set to avoid damage to the patient's lungs during delivery of the set tidal volume); 3) time-cycled ventilators set tidal volume by setting the inspiratory time and flow rate; and 4) flow cycled ventilators, where the inspiratory flow is terminated when the inspiratory flow rate drops below a specific level. The most common commercial modes of mechanical ventilation both provide a specified number of breaths per minute (BPM) and are 1) synchronized intermittent mandatory ventilation (SIMV) where patients can take additional breaths over the set rate and 2) assist control (AC) that uses triggering so that if the patient makes an effort to breathe, it helps them, and if not, it maintains the set rate. These modes can be used alone or in concert with 1) continuous positive airway pressure (CPAP), which uses a high-pressure reservoir and constant flow of gas that exceeds the patient's needs; 2) positive end-expiratory pressure (PEEP), which increases the residual reserve capacity and allows for many alveoli and small airways to remain open that would otherwise close off; or 3) pressure support ventilation (PSV), which adjusts the pressure on the fly as the patient breathes to maintain a preset inspiratory pressure.
For those designing open source ventilators using any of those modes and methods, there is a good base of established literature to draw upon. unless it both provides all of the source (as detailed above) to replicate it as well as shares it with a license that protect others' freedoms to make or use it. There are some flawed uses of this term from two types of designers. The first type consists of designers claiming they have open source projects before they have shared the code. This is the most rampant in the current ventilator design community with many pretty renderings and high-production value videos with nothing of technical value behind them (i.e. there is no source to replicate the machine available). Most of these designers may have good intentions but the source code may never materialize. Perhaps the most highly publicized case with a good ending was of Medtronic, a large commercial ventilator company, which first announced an open ventilator project on 3-29-2020, but did not release the CAD, BOM, software, etc. to actually fabricate it. Medtronic has now released these documents under a permissive license for their Puritan Bennett 560 ventilator, which already has been commercialized ($10,000 and first introduced 10 years ago) and received FDA approval. Although these design files have been accessed over 90,000 times, this system is designed for mass manufacturing and will likely only be manufactured in that context. All ventilators made from the designs must be labeled with a warning noting that it was built in response to COVID-19, and is only to be used to address this pandemic. Existing peer-reviewed literature The peer-reviewed literature itself is currently limited, but there has been some research on low-cost ventilation, even if the source is not available. First, a field portable ventilator system for domestic and military emergency medical response has been conceptually designed, but does not include enough information to construct it (e.g. the software was written in assembly language and not shared) 97 . This article does contain design considerations that may be useful for open source designers.
A new, compact and low-cost mask respirator concept has been developed and prototyped successfully 98 . The blower unit was able to provide adequate ventilation to the test lungs. In addition, the integrated sensor for airway pressure was able to detect airway occlusion and leakages. It is a relatively low-power device and could be operated wirelessly with batteries. It provides a cross-sectional view of the blower unit and some details, but again, not enough to be considered full open hardware or to be easily replicated. It should be noted, however, that many of the components are within RepRap-class 3-D printing capabilities.
In addition, research has been undertaken on a pre-stage public access ventilator (PAV) 99 . The PAV is made up of several low-cost technologies including a self-designed turbine and a range of sensors for differential pressure, flow, F i O 2 , F i CO 2 and three-axis acceleration measurements. The PAV was tested under three conditions to show that it was adequate for an automatic emergency system: 1) pressure-controlled ventilation (PCV), 2) PCV with controlled leakage and 3) PCV with simulated airway occlusion. The PAV was tested for and showed effective ventilation for tidal volume, breathing frequency and inspiratory pressure. Similarly, there has been a proposal to replace artificial manual breathing unit (AMBU) bags with electric blowers to act as emergency ventilators 100.
In contrast, another approach is to build a low-cost ventilator utilizing an AMBU bag that is not based on constant blower use 101 . The study by Mukaram Shahid showed the AMBU setup was able to perform all the functions of a conventional commercial ventilator for a far lower cost (<$100US excluding labor). The automated AMBU device was able to adjust the breathing rate and the volume of the air, which is comparable to older ventilators. However, it was also able to regulate the inspiration to expiration ratio and PEEP rate. Shahid's system comes with two modes: 1) mandatory ventilation (as in older models) and 2) assisted ventilation (as with most current systems). Thus, the medical personnel can choose to use either the built-in triggering mechanism (assist boosted mode), which alters the respiration pattern once it detects a change in air pressure, or set a time interval for the respiration pattern. The article contains pictures, an electric schematic, a control loop diagram, and very basic results. Again, this can be used as starting point, but there is not enough shared to replicate in the open hardware fashion.
Next, a low-cost ($420 prototype) portable mechanical ventilator was designed and prototyped that delivers breaths by compressing a conventional bag-valve mask (BVM) with a pivoting camactuated arm pushed by an electric motor 102 . This eliminates the need for a person pushing on the BVM, which is generally viewed as only a short-term solution. This system uses knobs to determine the tidal volume appropriate to the patient (usually 6-8 mL/kg of ideal body weight), adjustable BPM of 5-30, and inhalation to exhalation time ratio options of 1:2, 1:3 and 1:4 and a minimum respiratory rate 103 . This design is run with an open source Arduino micro-controller 104 and the article provides enough details to be used as a guide for others to build a similar device, but not the full plans, code, etc. needed to qualify as an open source hardware device.
One of the most relevant designs is a pneumatic ventilator specifically designed for pandemics, which has a low oxygen consumption 105  The results of this study support the potential for mass distributed production of a low-cost, gas-powered, volume-controlled ventilator with a low oxygen consumption (anywhere with oxygen at 2-4 bar). The designs could alternatively be operated on hospital compressed air. The single use, self-inflating bellows system prevents cross contamination among patients. In addition, the system possessed one-way and safety overpressure valves, which could be incorporated into other designs. The designs are in part supplied including basic principle schematics, an example BOM, but falls far short of what is expected for a complete open hardware design.
A large multidisciplinary and international team has just published (currently accepted, available in pre typesetting form) in a study on a low-cost, easy-to-build non-invasive pressure support ventilator meant for under-resourced regions 106 . The design is based upon using off-the-shelf components and is comprised of an open source Arduino Nano for control, high pressure blower and two pressure transducers. It was bench-marked against commercial systems. Their supplementary material also covers the testing with healthy volunteers, but more importantly, has the basic layout of the device, PCB and circuit schematics including source files, a BOM, STLs for the 3-D printable case, description of the algorithm and the Arduino ino file, and a user manual. This device's source is available and would represent a method to fabricate a ventilator for <$75, which has already been vetted by medical professionals. There are several interesting points about the approach used 106 . First, Garmendia et al. took the non-invasive medical approach, which is particularly well suited for both low-income countries 107 and also perhaps during pandemics where even the wealthiest nation's medical systems are strained. By focusing on off-the-shelf components their design could be easily replicated. In 'normal times', this approach is second only to systems that can be completely digitally fabricated with local resources. In pandemic situations, it exposes why it is important to have many such designs, as the global supply chains have been disrupted 108-110 . Normally, in the U.S. to replicate Garmendia et al.'s design based on the documentation provided would only be expected to take a few days. With the disruption, numerous makers have been having trouble sourcing supplies in the U.S., and the lowestcost blower following the Garmendia et al. design has an estimated shipping of 8-18 days on 4-28-2020 in the U.S. There are alternatives for providing this function (both suppliers and devices), which is why it is important to have a 'diversity of solutions' 62 with as many alternative suppliers, components and possibly even digitally manufacturable parts as possible (e.g. there are already several 3-D printable centrifugal blowers developed, which would demand future work for this application). Lastly, this design did not appear to have a license associated with it being a purely medical science publication. Even the Arduino code, which did have an author information for help, did not contain any license. This could hamper rapid deployment in some contexts as not explicitly indicating a license declares an implicit copyright without explaining how others could use the code 111,112 .
There are also completely different approaches to the design of a ventilator, such as the high-frequency oscillatory ventilator 113 , but only basic design schematics and preliminary testing is provided. Thus, within the peer-reviewed literature, most of the quasi-appropriate ventilator devices use a standard ventilation bag that is cyclically compressed by either an electromechanic or pneumatic setup and controlled by a microcontroller. Fortunately, the most complicated part of these designs is the controls, which is made accessible by the maturation of Arduino-based microcontrollers that can actuate and sense over a wide array of accessible and already-developed technologies (e.g. code libraries are available). It should be noted that most of the lowcost options in the literature used the bag approach, but that modern commercial ventilators are generally not manufactured with bags, bellows or pistons due to performance concerns. These concerns may be overcome by the nature of a pandemic, as well as by replacing low-cost components during failure, but this does indicate failure detection is warranted and certainly preferred in an open source ventilator design.
Open source ventilator designs shared on the web There are a number of proprietary commercial low-cost products like the Pumani bubbleCPAP for infants, D-box or One Breath Ventilators (not yet for sale), which could be used to relieve some of the demand for conventional ventilators. Rather than attempting to conduct a market review of such devices, however, because presumably hospitals facing a shortage of ventilators would already consider all commercially-available and regulated/ approved systems, this section will investigate the growing body of knowledge to help makers develop open source ventilators as well as the preliminary designs. This section was largely supported by information gathering of the rapidly evolving open source Internet communities such as Project Open Air, which is a group of "Helpful Engineers" on the platform Just One Giant Lab. They have congregated to help in the COVID-19 pandemic by developing open source solutions and of most relevance to this study, on a project specifically on the development of open source ventilators. Their documentation and information is freely available. Although just starting, as of 17 March 2020, they have over 2,500 registered volunteers and over 9,000 on their Slack team and by the beginning of April numbered over 15,000. In addition to an offset ventilator, in their first round of project proposals, they have prioritized oxygen concentrators and PPE as their top priority projects. In addition, their future work will focus on tube connectors and building a database for local manufacturers able to produce hardware with high score in reviews.
There are other teams working on the development of open source ventilators. Their progress is rapid and there appears to be more groups being formed and joining regularly to address needs in their communities. Robert Read et al., have been attempting to stay on top of these in a COVID-19 Ventilator Projects and Resources with FAQs available on GitHub. This resource contains a color-coded spreadsheet of the various projects and scores them on openness, buildability, community support, functional testing, reliability, COVID-19 suitability, and clinical friendliness and then ranks them by their average score. One can argue with the ranking, but the value of the resource is clear and all projects when they have obtained a reasonable level of development should ask to be evaluated. In addition, the spreadsheet has projects broken down into modular components whenever possible including drivers, monitors, flow sensors, display, oxygen blending and valves. To assist these efforts the UK government has issued guidelines.
There are several approaches being attempted in the open ventilator community including pumps, pressure regulators, bellows, pneumatic systems, screw compressors, servo gas modules, fans, blowers, fluid based, cuirass (negative pressure/iron lung), and pistons. The most favored by both the academic literature as well as the maker community is just to use manual ventilators -BVMs/ AMBU bags. There are many commercial suppliers available and there is very preliminary documentation for open source manual ventilation for the developing world 115,116 . Although, in theory, purely manual ventilation could work to provide ventilation for patients over long periods, there is a real concern of both the availability of the needed man-power, as well as the continued exposure of the laborer. In addition, using a bagvalve mask may increase aerosolization of virus, and in general medical staff are not supposed to bag mask before intubation due to that risk. Many of the open source designs rely on this BVMs/AMBU bags approach where one automates the manual squeezing. It only needs an exhaust system and PEEP valve. Students at Rice University have also created an automated bagvalve mask device that fits around a normal BVM using a dual rack and pinion design with a servo motor that continuously operates (open/close) squeezing the bag a specific amount to supply air. Rice provides a full non-peer-reviewed report, that is considerably richer in details than most of the others. It offers their design strategy, a partial BOM, basic testing, the source code as well as a summary of the standards and regulations necessary to go to market. Unfortunately, in their preliminary testing, the servo motor failed after only 11 hours of service and Rice is withholding the full CAD designs and results. To overcome the limitations of both the MIT and Rice designs, a group in Ireland formed and is moving along with full open source documentation of OpenLung on GitLab. The German language DIY-Beatmungsgerät project They are on their fifth iteration as of this writing based on the surrounding low-cost BVM/AMBU bag concept discussed above. Another project building off the MIT design is DIY Ventilators. Finally, the open hardware OxyGEN project is also using automated AMBU approach and although at the preliminary stages their 3-D and MATLAB design files are hosted openly on GitHub. The Oxy-GEN current system is under production in Spain.
Makers are also considering other types of non-invasive ventilators (NIV) such as those based CPAP (an alternative to PEEP), which is a form of positive airway pressure ventilator that applies mild air pressure on a continuous basis. A 3-D printed CPAP fan has been designed and tested as a blower and the design files (AutoDesk Fusion 360) and STLs are freely available. Another approach is to turn a commercial CPAP machine into a ventilator currently under development on GitHub by Lee. Lee built the system around an Arduino nano and has performed very basic tests to it that show that it provides enough pressure for a ventilator used on COVID-19 patients; however, there is not nearly enough information to recommend it for medical use.
In addition, there are bi-level positive airway pressure (BiPAP) machines that are commonly used at home to treat sleep apnea and lung diseases as they decrease the effort of breathing by changing the pressure for inhalation and exhalation. Homeuse BiPAPs could be used in place of hospital NIVs, but care would need to be taken because poor interfaces could generate viral aerosols 117 . Negative pressure ventilation (iron lung) overcomes this problem, helping lung function by pulling from the outside (there has been some development on Appropedia In the review of Internet-reported ventilators, it is somewhat disappointing that many of the most promising designs do not share their source code. Designers that do not share their source making their projects functionally non-replicable. A current representative example would be the Utah-Stanford Ventilator vent4us, which although looking promising and using an innovative a linear actuator-driven pinch valve-based implementation has only indicated they will release their designs in the open source domain, but has not (as of 4-28-2020), despite preliminary evaluation and submission to peer review 121 . In fact, in many cases, little more than a picture or video are available (e.g. Drexel University's Dragon Ventilator Project as of 4-28-2020). The newer projects do tend to be following better documentation protocols. Unfortunately, despite the many promising approaches in the maker community, the one problem hat the vast majority of the current partial designs have in common is that there is not nearly enough information available about their performance to recommend them for medical use.

Future work needed
It is clear from this review of the peer-reviewed, gray and open web literature on open source ventilators, that there is considerably more work to do. The tested and peer-reviewed systems lacked complete documentation and the open systems that were documented appropriately were either at the very early stages of design (sometimes without even a prototype) and were essentially only basically tested (and some were not tested at all).
With the considerably larger motivation of an ongoing pandemic, it is assumed that these projects will garner more resources and members (as is happening with the Open Air Project) to reach a critical mass to make significant progress to reach a functional and replicable system. Although the motivation of working during a pandemic on a device that may save your life is high, the access to resources, however, is far from optimal. Already, many locations throughout the world are essentially forcing citizens to shelter-in-place, which restricts access to government and university labs, as well as to makerspaces and fab labs. In addition, some areas of the world are suffering from supply disruptions and shipping challenges. This perhaps underscores the importance of developing open source hardware for disasters before the disaster strikes. Future work is needed to develop policies and funding mechanisms for such work as it appears rational to make a small investment in developing and sharing the designs for any critical hardware.
For those planning to work on (or who are already working on) the development of open source ventilators one of the primary challenges is to determine when to share your designs. People are literally dying from lack of ventilators and it is hard for designers not to feel responsible if they are reasonably confident a preliminary device design would possibly prevent those deaths if shared. Many makers follow this belief and often aggressively share their content before there is any evidence that it works. At the same time, well intentioned engineers and designers can have their work mischaracterized and promoted before it is documented by overly aggressive public relations outfits at both companies and universities, which has greatly added to the clutter in this space. On the other hand, as these are medical devices, which literally can mean life and death for a patient, it is reasonable to want to follow the conventional hardware developers' method: wait to release it until it has been fully tested. In addition, the effort and time it takes to do full documentation correctly may also appear to be lower priority than the making, prototyping and testing of the device itself. However, as Bowman 122 points out the "the intent to share a design in the future misses the myriad benefits of open hardware -in terms of scrutiny, feedback, and improvements from the community. It also stifles the development of a community around the design, and there are many cases of promised openness never materialising. Another challenge with this approach is maintaining a proper level of sterility of devices fabricated using distributed means. Specifically, for the FFF-based 3-D printing parts, it has been reported that the prints are sterile at the time of print 129 . If not kept in a sterile environment, however, they could quickly become biologically contaminated. One approach to deal with this is to use washing or a chemical bath. A relatively complete analysis of the chemical compatibility of commercial 3-D printed plastics is available 130 . If a specific polymer is needed that cannot be 3-D printed easily, it is possible to make molds in hightemperature plastics, such as polycarbonate, and then use lower temperature plastics to make disposable single use plastic parts 131 . Similarly, silicone molds can be made from a 3-D printed reverse mold and used in the same way 132 .
Even when more mature open source ventilator designs are available and can be safely manufactured by a distributed means, another area of critical future work is validation of these designs. In the medical sciences, open source devices like syringe pumps 133,134 are already established 135-138 and have been developed into sophisticated devices 139-144 . However, these devices are used in labs in general and not on people continually. For medical professionals to use an open source ventilator, they first must be convinced it will do no harm to them (or others) as well as to the patient. As COVID-19 was reported to spread via droplets, contact and natural aerosols from humanto-human, there has been a concern that high-risk aerosolproducing procedures may put medical personnel at high risk of nosocomial infections, which is a concern for some designs reviewed here 145  However, technical validation may not be enough. Medical hardware used on humans is also more complicated, as any studies involving humans needed to verify its functionality on people, need institutional review board approval and, if in regulated areas like the U.S., such a study would need an Investigational Device Exemption to allow for a non-FDA approved device to be used as part of a study. This is only a temporary approval and the full device would need actual FDA approval for legal deployment unless the laws are changed (or were temporarily suspended during a pandemic). These same regulatory roadblocks are in place in other nations, which has conventional ventilator manufacturers skeptical that even conventional manufacturers of other products (e.g. vacuum cleaner and automobile manufacturers are doing this now in the UK) could switch over to produce ventilators 158 . Clearly, this process is a problem during a pandemic. Both for the current situation and during potential future situations, there is a need to limit liability on the part of the designers, makers and users of such open source medical hardware 159 . One approach is for 'Good Samaritan' laws to be expanded to protect both the makers and designers of open source medical hardware 62 . Substantial future work is needed in this area. Finally, it should be pointed out that personnel and training can become limitations to deploying mass medical efforts, even if open source ventilators are available. So, future work is needed to create training materials and translate it into the languages spoken throughout the world as well.

Conclusions
There is clear technical potential for alleviating ventilator shortages during this and future pandemics using open source ventilator designs that can be rapidly fabricated using distributed manufacturing. The results of this review, however, found that the tested and peer-reviewed ventilator systems lacked complete documentation (with one recent exception) and that the current open systems that were documented were either at the very early stages of design or had undergone only early and rudimentary testing (although this is changing rapidly). With the considerably larger motivation of an ongoing pandemic, it is assumed these projects will garner greater attention and resources to make significant progress to reach a functional and easily replicated open References source ventilator system. There is a large amount of technical future work needed to move open source ventilators up to the level considered adequate for scientific-grade equipment and further work still to reach medical-grade hardware. Future work is needed to achieve the potential of this approach not only on the technical side, but also by developing policies, updating regulations and securing funding mechanisms for the development and testing of open source ventilators for both the current COVID19 pandemic, as well as for future pandemics and for everyday use in low-resource settings.

Data availability
No data are associated with this article.

Acknowledgments
The author would like to thank all of those currently affiliated with the open source ventilator communities and the associated open hardware and 3-D printing communities for their inputs in the cited references.

Introduction
The manuscript presents a review of open sources ventilators with particular emphasis towards applications in the COVID-19 pandemic during the early phases of 2020. Given the submission date of the manuscript, which has coincided with the emergence of the COVID-19 pandemic, the review is very topical, in addition to being thought provoking and insightful. COVID-19 disease is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), where the presenting symptoms of a patient are predominantly based upon respiratory ailments, requiring ventilation based equipment to treatment the most severe of cases. As the virus infiltrates a greater frequency of the general population, the availability of ventilation based products has rapidly become one of the primary causes for concern not only in developing, but in developed economics. Given the disruption of standard medical device supply chains during a pandemic, a viable alternative to meet the demand for ventilation equipment from healthcare providers is to turn to communities of designers, engineers, industrial specialists and knowledgeable maker enthusiasts to develop easily accessible, low-cost and open source alternatives to traditional devices. The manuscript is therefore a vital piece of documentation to assist researchers in global efforts to create ventilator alternatives. Equally, the manuscript does an exceptional job at highlighting the current state of the art in this area with balanced and considered conclusions throughout. I would therefore very much recommend this article for indexing, which will be of great interest to the scientific community and those seeking to develop their own open source solutions.

Recommendations and Thoughts
It is noted that the article is based on the premise of discussing ventilator technology which are deemed open source. As review manuscripts generally attract a wider target audience that scientific bodies of work, it would seem appropriate for the author to define what is meant by open source, for the benefit of unfamiliar readers. Indeed, one may see general scientific publications are open source information by virtue of the information being in the public domain. However, it is clear that the authors perception of this would require a deeper level of 'disclosure' of the innovations presented by various research groups to allow for ready duplication and adoption of such systems. This is starkly evident later in the manuscript during discussions of existing literature. Therefore, the distinction of what constitutes open source in this context should be explicitly defined, ideally within the introduction.
In light of the rapidly evolving nature of the pandemic and the volume of initiatives that are attempting to In light of the rapidly evolving nature of the pandemic and the volume of initiatives that are attempting to provide viable, often, open source solutions, there is likely to be some developments that have not been addressed in the review. Generally, the author has done an exemplary job of drawing the readers attention to many of the most topical and noteworthy examples. However, one would imagine as we approach the end of the pandemic period that there is likely to be a wealth of additional technologies which will have surfaced and so perhaps a follow up review may be justified, ideally inviting multiple authors involved in such projects to contribute. It is however noted that certain information within the manuscript has since evolved, specifically relating to the comment in the first paragraph of the introduction, where it is mentioned that '…….for a recent example, consider the fact that a manufacturer Upon threatened to sue a maker for 3-D printing life-saving valves in Italy for patent infringement….' further development of this story, the company had made official statement to clarify that they did not attempt to sue the party which made ventilator valve parts but had primarily withheld designs based upon medical device regulation. Although withholding designs during this particular circumstance when supply chain needs could not be met resulting in potential mortality of patients, there is an argument to disclose such information despite legal implications. However, this is very much different to the notion of the company suing the Italian firm. At the time of writing the article this would not have been known to the author, but given the controversial nature of the comment, I would request the author to reword this sentence to reflect the final outcome of this case study.
During the introduction when discussing the very many technologies available to the open source community, the discussion preferentially revolves around the use of rep-rap 3D printing, under the notion of digital fabrication technologies which have distributed manufacturing potential. This is a very important point to make by the author as distinctions are made as to why this approach would provide added value within the context of a pandemic. In particular, designs may be shared both at a national and international level using internet based data transfer, while leveraging manufacturing and technical capacity closer to the point of use. Such capacity has long been utilised by the open source community, providing strong resilience in instances when typical supply chains are disrupted, as would be the case during a pandemic. However, this discussion appears to be somewhat incomplete for readers who are unfamiliar with this approach of manufacturing. More specifically, it may be useful for future readers to hear some mention of other digitally driven distributed technologies, such as milling/CNC machining, laser engraving/etching and other digitally controlled tools. Indeed, several of these technologies would serve to reduce or eliminate the perceived limitations of 3D printing both in terms of manufacturable materials and speed of manufacturing. Please could the author include some additional discussion here to contextualise available options to the open source community with respect to digital and distributed manufacturing.
The author presents an eloquent attempt to discuss both the academic and non-academic ventilation systems, citing many interesting studies and raising most of the key facets of each respective technology. Following from the previous comments regarding the open source nature of a given study, it is highlighted that sadly many academic studies, though presenting some remarkable feats of engineering, simply do not provide sufficient information to allow for other researchers and experts to duplicate a respective ventilation system. This is indeed a shame and highlights perhaps a necessity for researchers to adopt a more open frame work of reporting academic findings and equally for reviewers to encourage and accept such ways of reporting. In light of major global health catastrophes such as pandemics, such openness within the literature may in fact prove advantageous to hasten innovations to tackle the detrimental effects on a given population.
The discussion on the current open source efforts appears to cover several of the major projects to the awareness of this reviewer, but more crucially provides a good cross section of important developmental aspects to inform the reader. As highlighted previously, given the rapidly evolving nature of the pandemic and the wiliness and passion of researchers and technical experts, new attempts to create ventilation and the wiliness and passion of researchers and technical experts, new attempts to create ventilation products are arising on a weekly, if not daily basis from groups around the world. Therefore, in this reviewer's opinion, to cover all such attempts would be impractical but also unnecessary given that the overlap of technical development with existing projects. As with the previous section, it would have been good to present annotated diagrams and a summary chart/table of the systems broken down into the primary attributes that fulfil the requirement for ventilation. In reading this section it was difficult to surmise how these efforts were truly moving towards a functional ventilation device. One of the more noteworthy examples from Rice University, with credible data to back the developments to date was sadly hindered by component failure after only a 11-hour evaluation period, which falls considerably short of a functional ventilator. Clearly, several if not all examples highlighted by the author raise several notes of caution toward open source design, namely the clarity and robustness of evaluation, which lacks the rigour of typical academic scrutiny in addition to the lack of standardisation of components which are suitable for purpose. Arguably, the author to varying degrees' highlights this by stating the case studies in question are either in the very preliminary stages of investigation, show a lack of relevant performance data or a Bill of Materials (BOM) which enables scrutiny of the components employed. I believe there is a missed opportunity of discussing these elements in greater depth, which is vitally required if open source ventilators are to truly enter mainstream acceptance and use. I would very much welcome further discussion on limitations with appropriate recommendations, which both do not stifle the breadth of design ideas by the community, but also provide substantiative guidance to direct those involved to be mindful of critical milestones and 'codes of best practise' during the journey from inception of idea to final working and 'usable' ventilator. For example, looking at open source efforts for Personal Protective Equipment (PPE) we see that the 3D Printing community rapidly converged towards the preferential use of PETG polymers for manufacturing, owing to the factors of mechanical stiffness providing a semi flexible yet robust part, and importantly biocompatibility for limited human contact and being food grade to allow ease of decontamination. Equally, strict protocols were developed to minimise contamination of parts during printing, handling and shipping to both reduce the spread of the virus within the supply chain and to follow some element of best practise, similar to constraints set by medical device companies upon their manufacturing procedures. It would be very useful to the open source community if the author would share their thoughts in more explicit detail to provide a template that could be built upon for future efforts, outlining best practise from initial design ideation to working prototype. Arguably, the imperative of this is much greater than with PPE given potential intrusive nature of ventilation systems providing a clear route to internal infection through the lungs.
One element that appears to be missed in the present review is an overview of current commercial systems and evaluation of their performance characteristics to be both efficacious and crucially to prevent unintended harm to a patient. The author does cite an existing review of commercial ventilators by Pham , and so it would be unnecessary to conduct a repeat of this work. However, what would be relevant in et al the present manuscript is a discussion focusing on the regulatory and quality assurance aspects and how these would align and differ in the approach of open source ventilation systems. Indeed, commercial devices must adhere to the very strictest regulatory scrutiny to be classified and used as a medical device, particularly given the invasive nature of their operation to either supplement or take complete control of a patients breathing. As such there are tightly regulated frameworks of 'fail safes' to ensure that every component used falls within acceptable usage limits, that construction of such devices follow strict regimes for assembly and minimisation of contamination, that tests are conducted to evaluate the working performance of each device, that there is a robust training and best practise usage protocol, amongst many other safety measures and supply chain demands. Arguably, one of the biggest reservations by the commercial sector regarding open source hardware for medical device technology are the lack of process control, quality assurances and regulations regarding technological development and best practise. It would be valuable for the readers to have some insight into what safety and evaluation procedures that are conducted and how such measures would be replicated in open sources systems in general terms, are conducted and how such measures would be replicated in open sources systems in general terms, highlighting key challenges which can be built upon for the future discussion.
The final sections of the review offer an exceptional summary of the state of art in the field of open source ventilators, highlighting several challenges and opportunities in this space. The author rightly highlights the very early stages open source ventilations systems currently are at and that we should not expect these systems to be working within a clinical setting in the immediate future. Despite this, I for one feel optimistic given the current work that is underway, our ability to access a wealth of digital knowledge, alongside the availability of hardware and manufacturing resources at our disposal. Arguably, the open source community has never been better equipment to make positive impact on the world during the COVID-19 pandemic. It is also the opinion of the reviewer that there remains many more opportunities for both growth of the open source community and to leverage the expertise synergistically with other academic groups to more rapidly advance our preparedness for emergency situations.
Reflecting upon the manuscript, I could not help but feel there are differing schools of thought that of the traditionalist and the open source innovator. The traditionalist will operate with robust scientific rigour but will provide limited information scientific manuscripts, be open to patent and potentially restrict free flow of concepts and with respect to medical technology will strictly follow medical regulatory frameworks. The open source innovator is generally driven by an overwhelming sense of openness and transparency in their work, with the belief this will help proliferate and see ideas adopted faster for anyone's benefit, albeit on some occasions operate with a naivety towards regulatory and best practise aspects. Initial discussions by the author drew distinction between efforts of the academic and non-academic communities, which very loosely are comprised of these two types of innovators, with the exception of a few 'maverick' academic groups. However, there was little discussion of strategies as to how these two communities may come together in cooperation and bridge any perceived differences in thinking. Clearly the academic community prides its outcomes based on empirical evidence, the careful scrutiny of data, alongside objective design performance metrics, attributes that the authors clearly outlines as shortfalls of the general open source community. Conversely, the non-academic community provides a wealth of creativity, ingenuity, alongside technical prowess, finding often remarkable and highly efficacious solutions working with limited resources and minimal dependency of specific supply chains. I would therefore strongly welcome the author to add further discussion towards strategies on how both communities, which for the most part work independently, could align agendas to realise opportunities that transcend the sum of the two parts. Indeed, I think it more critical for the open source community, based upon the reflection of the author, to be more engaged in traditional scientific process and to incorporate this into their thinking to hasten product development for evaluation.
One element that gave me considerable food for thought, was the notion that developing nations have a perceived advantage during global health emergencies due to their more relaxed legal and regulatory frameworks to deploy open source ventilator systems. I am not entirely convinced that this is the case and indeed the author makes several valid arguments to the contrary, mainly that the technology is not significantly mature to function as intended and without consequence. Such suggestions can indeed have several unpleasant connotations from a legal and ethical standpoint and so I would encourage the reviewer to consider an amendment to this comment. Ultimately, I believe it was not the authors intension to imply this given previous discussions, however the context of this point should be framed better.

Summary
Overall, despite the sombre theme of the review, the author has done an admirable job of bringing together all the relevant themes relating to open source ventilation systems. One of biggest take home messages from the review is how much potential exists with the open source community to provide cost effective, robust and timely medical device solutions, which may be far less susceptible to supply chain effective, robust and timely medical device solutions, which may be far less susceptible to supply chain disruption and leverage a greater capacity for localised fabrication using the distributed manufacturing model. This capacity can only be realised by continued development of existing open source projects, increased dialogue with academic groups to work collaboratively to validate and iteratively improve ventilation system concepts for maximum efficacy. Equally, there is a clear need for regulatory reform which appreciates the evolving circumstances during a global health crisis and could provide an alternative framework to leverage capacity outside of typical medical supply chains to supplement efforts on the ground, as and when appropriate. What this framework should look like is another debate entirely, but this article makes an elegant argument for the debate to be had.
Despite the infancy of open source ventilation systems, much potential exists and it is an exciting time for developers to continue their efforts towards working solution. It is exciting to see what may be a paradigm shift in how we perceive and operate globally in the medical device sector, particularly in light of the recent issues during the COVID-19 pandemic, which have decimated supply chains, while the shear volume of cases has put a drain on medical resources. The pandemic has already seen the use of open source designs, manufactured in a distributed manner, make impact to supplement shortfalls in PPE equipment. Could the same in time be true for ventilator technologies based on the balance of growing demand and available resources? Only time will tell. I reiterate the relevancy of the article by the author and the manuscript has been a very thought provoking document to read and digest. I would whole heartedly recommend this article for publication and encourage researcher and technologist in the field to draw inspiration from the insightful and thought provoking arguments outlined. I do welcome a time when the lessons we are learning during the pandemic lead to a more caring and equitable world for us all, and it feels from a technological standpoint, that open source innovation will be part of that story.

Additional minor points for consideration
The hyperlink for Ref 82 needs revising as it links to an error page.
One the first paragraph of page 4 'breadth' has mistakenly been used in place of 'breath'. Additionally there are some minor grammatical errors toward the end of this same paragraph that need amending, specifically the sentence starting with 'texts area available for the ……..' Generally speaking, it would have been very useful to contextualise much of the discussion with annotated diagrams of several key open source innovations, to give the reader a real feel for the types of devices in development. Indeed, this is typical of many academic based reviews of the scientific and gray literature. Unfortunately, this has not been the case in the present manuscript and I would invite the author to consider such an amendment.

Is the review written in accessible language? Yes
Are the conclusions drawn appropriate in the context of the current research literature? Yes No competing interests were disclosed.

Competing Interests:
No competing interests were disclosed. I have removed the now incorrect example of the manufacturer suing over reverse engineered valves. Although it may be interesting to note I contacted the individual who reversed engineered them before the initial article was published and I have yet to receive the STL files.
I share your frustration with the current lack of appropriate sharing. I have included a more detailed review of the existing designs -but as you point out it is impractical to do everything in such a rapidly changing field where websites are being updated hourly or faster. I have declined to develop diagrams -even if I could find appropriately licensed images as a summary because of this rapidly changing nature. Particularly in mechanical designs because those appear to be the most often altered. In addition, following your recommendation I have attempted to provide some best practices in terms of 'when to share' in the discussion. This is meant to bridge the gap between the traditionalist and maker philosophies you discuss.
After considering your points and those of the other reviewers about the advantages of less developed regulatory systems I have simply removed this entirely.
I have corrected the minor mistakes you pointed out -thank you for finding them.
Lastly, I don't think that a complete review of all the legal hurdles and regulatory framework needed in this technical space can be done here -it needs a completely separate review. We have a major problem here because even some of the standards are not available. When an artificial lung company contacted their customers on my behalf to get a testing protocol they referred me to a long list of ASTM standards which I was not able to acquire through my relatively-well-resourced University library. I purchased the first one and was disappointed to find that it mostly contained references to other standards and a shocking dearth of useful technical information. As there is a these devices have the same high standard level as those approved for use in developed countries (e.g. CE / FDA marks). But it is also important that until the industry can provide such medical devices at affordable prices for LMICs, patients are not deprived of life-saving therapies. Finding a balance on this ethical-legal issue is difficult but fundamental.
The review of peer reviewed articles is interesting and does a good job of rating the different solutions in terms of openness; it is disappointing that these articles don't generally give sufficient information to reproduce the ventilator, but also unsurprising. This lends a great deal of weight to the current move towards more openness in science, where protocols, data, and schematics can be shared in data archives along with papers -but of course that's rarely done retrospectively.
The review of "internet and gray literature" seems objective and reasonable to me, and while such a review cannot possibly stay exhaustive given the frequency with which such projects are appearing, it does seem to cover many of the projects I've heard of. More important than an exhaustive list, however, is the discussion of the common issues to most of the DIY projects -the need for careful testing, quality control, and proper authorisation. Most discussions have focused only on technical validation -but as the author rightly points out, this is not the only way medical devices must be assessed. At least as pressing as the technical challenge is the difficulty of getting new suppliers and new devices through a quality assurance process that gives medical professionals the confidence that they can safely use said devices.
Openness is an important, and often surprisingly contentious, issue. Of the projects that are discussed, only relatively few make available complete designs for their solution. This is particularly surprising in the case of some projects from high-profile institutions that have already been widely reported in the media as "open" while not yet having released any designs. The commonly-accepted practice in open software is that complete designs, including source code and documentation, are made available to the public, and that a project is not considered open until this happens. Similar norms are being established for open hardware projects, supported by organisations such as OSHWA and GOSH.
Given the safety-critical nature of a ventilator, it's reasonable to be reluctant to release untested designs out of a desire to be responsible. Given the time-critical situation, sharing documentation and designs may also be considered lower priority than product development. However, the intent to share a design in the future misses the myriad benefits of open hardware -in terms of scrutiny, feedback, and improvements from the community. It also stifles the development of a community around the design, and there are many cases of promised openness never materialising. My own view is that projects ought not to claim openness until their designs are publicly available under an appropriate license, but there are definitely valid ethical and practical concerns here, and I would welcome an open debate on the best way forward.
The one statement in the article that I'm slightly troubled by is the suggestion that developing countries may be at an advantage due to their less robust regulatory systems. Firstly, while it is true that many countries in the Global South do have less formal economies, their regulations are often very tightly aligned with those in richer nations -for example, the Tanzanian medical device regulations closely mirror those used in the EU. If different standards are adhered to, it may be because the regulations are not implemented fully, rather than because the government has intentionally applied lower standards. Also, the better-resourced regulatory bodies in rich nations are more able to accelerate the process of approval if needed; it is not clear to me that a medical device would clear the bureaucratic hurdles and achieve approval any faster in a developing country, indeed the process can be much slower. It is also a very thorny ethical issue to trial medical interventions in the Global South that would not pass ethical scrutiny in richer nations, particularly as the interventions are often being proposed by people from said richer nations. I don't think the author is suggesting this, but I do feel it's a point worth highlighting. While there is often an argument made that low quality medical supplies may be better than nothing, it is also reasonable to expect that developers of technology shouldn't do anything to citizens of Low and Middle-Income Countries (LMICs) that they wouldn't do to patients in their own nation. Indeed, most ethical review panels in the UK apply exactly this criterion.
The challenge of creating a safety-critical medical device that can be produced in a distributed manner is significant, and I think the article reflects this. I could not agree more with the statement that "technical validation may not be enough" and would probably go further, to say that technical validation alone is not sufficient to ensure patient safety. While many open ventilator projects now exist and have gathered impressive numbers of volunteers, there remains a significant global challenge to enable such projects to be regulated appropriately, either in the current crisis or longer-term. The existing system of medical device regulation is slow, expensive, and conservative; while this conservatism has its roots in the entirely reasonable desire to prevent harm to patients, the way the system is implemented makes it extremely difficult to certify a medical device without the resources of a large company. Reform of these regulatory systems could enable a more agile approach to the design and manufacture of safety-critical components, but a satisfactory supply chain will also require significantly more quality management than is present in a typical "maker space" run by volunteers, hobbyists, or even experienced engineers. Questions around training and liability are also of paramount importance; while litigation against volunteers acting in good faith seems unduly harsh, there must be accountability in the supply chain of medical devices. Otherwise, we push responsibility onto the clinical staff using uncertified equipment, which adds a crippling burden to front-line staff who are already working at the limit of their capacity.
Overall, I think it's right to keep an optimistic tone, while acknowledging the obvious difficulties associated with the current challenge. It's likely that, while there are many 3D printers available around the world, formal structures that do not yet exist will be needed to enable them to be fully employed to solve supply issues in this and future crises. Whether or not it is possible to make use of community designed and built ventilators in the coming months, I look forward to a world where critical supplies can be designed and produced openly for the common good. If we take the opportunity to put LMICs on a more equitable footing with respect to richer nations, the future may be more inclusive, as well as more resilient.

Are all factual statements correct and adequately supported by citations? Yes
Is the review written in accessible language?