Modular Design in The Ventilator Crisis: Why It Matters
This article is meant to be read by the engineers and NGOs communally attempting to address the world-wide shortfall of ventilators caused by the COVID-19 pandemic, or those facing similar future emergency scenarios.
In the last month the engineering community as a whole seems to have become a humanitarian engineering community, building upon the foundations laid by non-profits such as HelpfulEngineering.org and Engineers Without Borders. The Maker movement has succeeded in providing personal protective equipment (PPE) to medical workers around the world. Making a ventilator is two orders of magnitude more difficult than making a mask. At the time of this writing, powerful, competent firms such as Tesla, Dyson, Ford and GM have begun making ventilators. They may meet the need in the developed world and perhaps even globally, but preparedness is better than complacency.
A new pandemic is not a surprise; it was alway to be expected. As such we believe that the approach we describe below is valuable for ongoing efforts by the open-source community in the current design of ventilators, but can also be extended to any future crisis scenario for which the open-source community is developing highly specialized equipment. At its most successful, a modular approach to the challenges posed by the COVID-19 pandemic would result in not only a working ventilator but an entire ventilator ecosystem: a variety of modules which could be assembled to meet different healthcare provider needs, work around supply chain disruptions, or utilize available equipment or components.
In this article we’ll build the case for why the global open-source community should adopt a modular approach to the development of ventilators. This approach both (1) enables a necessary level of flexibility in medical treatment by healthcare professionals, as well as (2) accommodates local or broadscale disruptions in supply chain. As we’ll see, modular design also affords increased resilience, parallel design workflows, repurposability, and openness, all of which are advantages in highly variable crisis scenarios.
Flexibility of Treatment
It is not the place of engineers to tell doctors how to treat patients. It can be the place of engineers to give doctors every tool possible in sufficient quantities that they can make decisions as free from scarcity as possible. Particularly in the developing world, scarcity may limit the way patients are treated, thus producing far worse outcomes. In all cases and everywhere, we know that keeping patients strong through supportive care, in every stage of disease severity, will improve outcomes. Respiratory support can occur at many levels.
At the time of this writing, doctors are still discovering the best way to treat those stricken by the most severe symptoms of COVID-19, which is Acute Respiratory Distress Syndrome (ARDS), or at least similar to ARDS. As they learn, treatment regimes will change and improve.
Medical oxygen is critically important. It can be combined with CPAP, BPAP, non-invasive ventilation, or, when necessary, invasive ventilation via intubation with an endotracheal tube. All of these forms of ventilation have much in common; in fact they are thought of as modes of ventilation. Of course, ventilators have been specialized for these specific uses in the past when demand was predictable. Such specialization does not serve us well in the present crisis, and may not serve us well in future pandemics, which may come sooner than we think.
The engineering community must therefore turn towards flexibility. The most important form of flexibility is flexibility of treatment: All doctors everywhere should be given the best tools we can muster for each stage of disease progression.
Dr Erich Schulz, MBBS, FANZCA, has quickly written evolving essential reading for engineers on this topic broken into four short volumes: “A brief for engineers, by a doctor, on hacking a ventilator for surge capacity in Covid19 patients.” He lays out a strong case that without sensors to detect and support spontaneous breaths, pandemic ventilators will have narrow clinical applicability to passive (heavily sedated) severely ill patients that currently have high mortality under the best of circumstances. A large supply of pandemic ventilators will be far more usefully deployable if they can support both actively and passively breathing patients. We believe that modularity will support those features that will make them more deployable.
Local Shortfalls and Supply Chain Shortfalls
The need for mechanical ventilators in this pandemic is unpredictable, but we can say with certainty:
- There will be local shortages even if worldwide demand is met.
- When a health care system, including ventilators, is overwhelmed, outcomes will be worse. In other words, more people will die.
Disruptions in the supply chain of essential parts may hamper even the giant automotive firms. Sensirion is a major manufacturer of flow sensors with a reputation as a good corporate citizen. At this writing, they are attempting to fulfill six times their usual annual orders for flow sensors in two quarters. We hope they succeed and meet the global demand.
One can easily imagine a worst-case scenario, however, in which such a critical part cannot be produced quickly. Or, imagine that the part may have been made in a nation suddenly at war or in crisis. One can even imagine such a part being withheld from export for political leverage. The shortfall of flow sensors illustrates an important point: modern sophisticated ventilators are highly specialized devices which cannot be produced at enormous scales quickly enough for us to rely on them. In a worst-case scenario, or for those who cannot afford them, there will not be enough.
Supply Chain Resilience
Ideally, we would have supply chain resilience. Nobody can make a Sensirion flow sensor better than Sensirion; but nobody should die because they cannot get one. We, as engineers, should construct a system which is resilient to supply chain disruption. We should not aim for perfect substitutability, but we should aim for graceful degradation. This is a term of art which can be put more colloquially: the system should bend, not break.
Supply chain resilience must be considered at every scale. It is useful for someone in the most remote region to be able to build, with extraordinary labor and skill, one ventilator without any access to a supply chain of parts. However, it is more useful to be able to build 10,000 in the presence of supply chain disruption. If the COVID-19 pandemic is uncontained just within the crowded nation of India and if only one in a thousand people need a ventilator (an extraordinarily conservative estimate), then one million ventilators are needed in India alone. We can hope, particularly with other curve-flattening interventions in place, that not all patients would require treatment at the same time. Unfortunately COVID-19 makes the most affected individuals sick for a long time, usually about a month.
The only way we can see to achieve supply chain resilience on this scale is through modularity.
Modularity = Resilience
By modularity, we mean that the system consists of a system of parts, called modules, which can be interchanged in a LEGO® brick -like fashion. The concept is of course familiar. It may even be present within the product lines of major firms, but until now they have had little incentive to reveal, or open that modularity.
If a flow sense is a module, it can be replaced with another sensor, perhaps one that is not quite as good, or one that is more or less expensive. If the entire sensing system is a module, it can be replaced with another sensing module made from available parts.
Modularity lets us design for an open possibility space: for an uncertain future in which supply chain lapses might affect the availability of necessary components, but also a world where local healthcare provider needs or specific implementations of ventilator device design may vary.
To some extent, modularity is the enemy of optimization. Modularity demands separation by rigid standards and protocols that precludes optimization through tight integration. Specialization has historically served us well. Highly optimized and specialized industrial designs are good. Having a modular, resilient alternative is also good. That is what we must build now to make sure the most people possible stay alive in the wake of the COVID-19 pandemic.
Just by standardizing the way individual pieces fit together, LEGO bricks open up countless possibilities for complex structures and the design of new components. The global open-source community would benefit similarly from a modularized approach to emergency ventilator design.
Parallel = Fast
Modular design also lends us an advantage in the timeline of design and development of crisis interventions. By effectively modularizing the problem space, individual project teams are each able to concentrate their efforts on one piece of the engineering puzzle, and thus all of the research, development, user and lab testing, and all other stages necessary for each of these designs happens concurrently. Since the scope of each module is simplified relative to an overall ventilator design, this development and testing process would likely be expedited and can be more precisely tailored to the nature of that specific component’s task.
Under this approach, as individual modules are refined to a level of operability, the entire community benefits. If by contrast an individual module fails or returns to the drawing board, it has minimal impact on the rest of the ecosystem. Early on in the open source community response to the COVID-19 pandemic we saw many concerned individuals pin their hopes on the eventual release of a handful of frontrunner full ventilator design projects, or on their own team’s efforts. The creators of those projects, understandably, were not comfortable releasing their schematics and software until the ventilator design had been experimentally validated. However, a modular , open approach to design and validation would have better benefited the entire community, allow many teams to all go faster and reuse each other’s work.
Software engineers will be familiar with the notion of parallel processing. This is parallel processing at human scale. The ventilator design development is already happening concurrently by simple nature of the fact that we are all working on the same problem at the same time, but as long as each project is a standalone unit, we don’t reap the ecosystem-level benefits of a parallelized approach to that process. A modular design strategy does.
Softening = Versatility
For the last 50 years, machines and parts of machines have been getting more electronic control. One could say that they have been getting smarter. We prefer to say they have been getting softer, in the sense of software as opposed to hardware. As the machines and even tiny, inexpensive parts of machines have gotten softer, they have become more versatile. (This is why it may be possible to “jailbreak” a given CPAP machine to turn it into a BPAP machine or simple invasive ventilator.) In 2020 we talk about the Internet of Things (IoT): the widespread intercommunication of devices which in the past were just things. Now they have become sensing, internetworked things. This is an opportunity.
This softness, or reprogrammability, allows repurposing. It also allows modularity. Two motors may not be the same: they may not provide the same force, or they may not have the same duty-cycle. However, if they can both be programmed to do the same thing, these differences can be made irrelevant in an emergency pandemic ventilator. The motor can be thought of as a module that supplies rotation. One particular motor may be preferable, but both are serviceable. They both will keep you alive.
Openness = Confidence
However, to achieve this modularity, parts and modules must be open. You cannot have confidence in replacing a part if you cannot see inside it. Sight is here used as a metaphor for being able to understand, examine, and test the part. You cannot have confidence in that which you cannot verify, and you cannot verify what you cannot examine.
Regulatory agencies should not trust any single authority that a design or modified design is safe; they should rather demand transparency of design and obtain peer review or third-party review. This is only possible if designs and software are publicly and clearly documented — in a word, open. Because ventilators are life-critical medical equipment, any rapidly manufactured ventilator system must be open and independently constructable and verifiable.
Hot Swap is the Ultimate Resilience
The ultimate resilience is the ability to do a “hot swap” of components while the machine is in operation. This may seem unattainable, but in fact the proliferation of AmbuBag designs which do not specify the precise bag very clearly show it is possible, even a necessary design goal, in the current crisis.
A modular approach to open-source ventilator development will require the development of new standards, among other innovations.
In this article we argue why the global open-source community should adopt a modular approach to the development of specialized emergency equipment such as medical ventilators, defining four mottoes to encapsulate this approach (“Modularity = Resilience”, “Parallel = Fast,” “Softening = Versatility”, and “Openness = Confidence”). A follow-on article, Modular Design of Pandemic Ventilators, that will be maintained as an open-source living document, provides more practical detail of how the open-source community could effectively adopt a modular approach to design and development in emergency scenarios, in particular that of ventilator design to address the COVID-19 pandemic. We encourage your suggestions made within that document.