Yesterday NASA’s Jet Propulsion Laboratory announced that their ventilator design has received Emergency Use Authorization from the US Food and Drug Administration. This paves the way for the design to be manufactured for use in the treatment of COVID-19 patients.
JPL, which is tightly partnered with the California Institute of Technology, designed the ventilator for rapid manufacturing to meet the current need for respiratory tools made scarce by the pandemic. The design process took only 37 days and was submitted for FDA approval around April 23rd. They call it VITAL — Ventilator Intervention Technology Accessible Locally — a nod to NASA’s proclivity for acronyms.
What’s Under the Hood? We Don’t Really Know
There is a website dedicated to the VITAL project but it is very light on hardware details, a topic we know you’re hungry for. I’m left to read between the lines. From the Notification of Licensing Opportunity (PDF) we learn that this is not equivalent to an ICU ventilator, but one to be used leading up to that need:
JPL’s ventilators are targeted to treat the segment of COVID-19 patients who are symptomatic and in need of pre-ICU respiratory ventilation.
Furthermore, the brochure (48 MB PDF) mentions that the system targets patients who are not improving with supplemental oxygen:
VITAL DEVICE INTENDED USE: […] Patient Unable to maintain adequate/safe oxygen level with nasal or mask oxygen […] Intubated, sedated
Hackaday reached out to JPL and confirmed that VITAL is not a design involving an Ambu bag — an approach I’ve been skeptical about in past writings — instead taking the more common blower-based approach. It is an invasive ventilation design, requiring patients to be intubated and medicated during respiration. It is capable of both reacting to the patient’s own breathing cues in an assisting role, as well as ensuring a minimum breathing cadence (the “back-up rate” setting shown on the control panel).
The user interface opts for 7-segment, bar-graph, and discreet LEDs with buttons for user input. This makes operating the device simpler, as features aren’t hidden behind several levels of menus as they might be with a touch-screen interface (hopefully ensuring respiratory professionals can quickly get up to speed with this model). It is designed for 20 days of continuous operation at 100% duty cycle. Interestingly the service life is just 4 months — I’m uncertain if this is an issue with wear and tear on the parts, or due to the difficulty of sanitization regimes for ventilators.
Where I have been most skeptical with all recently announced ventilator designs is the software. However, I know that NASA has a strictly codified and time tested approach to software verification for their flight hardware which in many cases is responsible for keeping humans alive in extreme environments. If this was followed in developing software for VITAL, that alleviates much of my concern.
Can We Build Them?
JPL has done the design work, but they do not have the manufacturing capacity to scale this up in a meaningful way. To that end, the design is currently offered under a free license to applicants selected on a case-by-case basis. Applications are due by May 4th, with selection to be concluded just a few days later on the 7th.
This means you can’t look at the design right now, and it’s not currently open source. There are a couple of ways to look at this tight control. First, this ventilator is designed for use with sedated patients monitored by respiratory professionals, and not something that should be attempted outside of the medical setting — controlling access to the design may be an effort to help ensure these aren’t built and used by untrained persons unnecessarily leading to injury or death. Second, the supply chain is what is currently causing a ventilator shortage so controlling the design may help keep the parts that are in shortest supply going to manufacturing efforts with the best chance of success — if it were released openly, people may begin snapping up as much of the bill of materials as possible causing further shortages.
Will We Share Them?
JPL confirmed to us that this design is not open source but is offered under a world-wide no-fee license administered by CalTech during the COVID-19 crisis. This is vital technology throughout the world and my hope is that it will be shared with any entity making a reasonable request, but only time will tell on how this plays out. I’m told that over 160 applications were received on Thursday alone.
That said, equally important right now is that in a world with a limited supply of available respirators, we should be sharing them. Hand in hand with mitigating the spread and resurgence of COVID-19, humanity must take an active role in procuring, maintaining, and sending these machines to any area in need.
Congratulations to the NASA/JPL team for the approved design, and to the FDA for working through the authorization process quickly. We hope the search for manufacturers will go well and that production rises to meet the need during this crisis.
Another reason for control, some designs for other things got emergency use authorisation or approval from other responsible bodies and were released, then there was a flood of “I improveded it !!!” derivatives being MADE. Not submitted for re-approval before release, being put in the “market” as it were and confusing the hell out of people trying to co-ordinate purchasing, distributed manufacturing or volunteer donation of such products, unsure what they’re actually getting.
Ah, this is a valid point I hadn’t considered. Thanks!
some peek “under the hood” can be seen in this video:
https://www.youtube.com/watch?v=NB7SdwkBqHU
Yeah, I scrutinized that one. They have the face panel off the thing at 1:35 but I think most of the guts are still missing at that point.
I would think the 7 seg displays are harder to source than a 4.3″ or 7″ LCD and an RPi/NanoPi/OrangePi. Maybe they wanted to avoid Linux/Android/Winemebedded, which is an admirable goal. There must be some processor. Does it have to run a medical/flight qualified RTOS like uC-OSII or III?
Why do you think 7 seg displays are uncommon? Thinking about my house, the washer, dryer, refrigerator, cable boxes, clocks, and thermostats all use 7 segment displays. Digikey has a stock of over 100,000 7 segment displays: https://www.digikey.com/products/en/optoelectronics/display-modules-led-character-and-numeric/92?FV=-8%7C92&quantity=0&ColumnSort=-1000009&page=1&k=7+segment&pageSize=25. They’re also easier to drive, have great contrast, and have no refresh time. Not to mention you don’t need to have special support for each type, you can pretty much swap them out with another brand after changing the pinout, the software’s the same—or make your own with a few LEDs. There’s no reason for a full OS, a simple RTOS is all you need. It would probably take a lot longer to get a full OS and/or an LCD approved.
What he said, and
7seg:
-cheap
-reliable
-bright/high contrast
-reliable
-large
An LCD will cost more to achieve the same brightness and contrast, doesn’t clearly locate the data (it is all in one small rectangle, rather than in well identified zones on the faceplate), takes boot, configure, and update time (probably not a big issue, but still), and are not as rugged in the wild.
… and if supply really, really, gets boned, you can 3D print or print moulds for housings to stuff practically any kind of discrete LED into to do the same job.
mmmmmm big no. As someone who works on small volume projects, that need to made quickly, LCD’s just suck.
Things are clearly labelled and not with abstract icons/GUI. They are just the way they should be.
+1, for such dedicated tools. This is not where you want LCD, with limite angle of vision, small characters, and no, no touchscreen!
Ah. I wonder how many patents were broken with this design. And also how easily the FDA’s objections were waived. I have experience with FDA applications, and I can tell you that an application takes multiple months. And 1 little mistake will make them blow off everything, and you can start all over again.
I do work in a company making medically approved equipment. And even though our equipment is approved by many different important institutions over the world, the FDA application is always the hardest ever. Maybe it’s because my company is not US-based.
Anyway. Great feelgood story. Even though there are many companies in the world making perfectly good and approved ventilators, and the only thing that they are missing is production capacity.
And production capacity is the one thing that NASA *didn’t* tackle with this ventilator.
Can’t make a medically approved ventilator without a medically approved production process.
This does not look cleanable in the way one expects in a hospital, though I suppose you can “gown it” with a plastic bag. By the way, is there a shortage of ventilators?
Is there a shortage of ventilators?
Good question. Here are two more that go with the first.
Will COVID-19 be prevalent until 70 percent of the population is infected?
Will there be a second wave?
Here’s the answer in light of all three questions:
Yes.
Well, it’s NASA… so it will be five years behind schedule due to testing red tape and be produced by a corporate welfare recipient like Boeing at cost-plus for three times what it should.
I work at JPL, but don’t speak for them here, nor was I involved in the design team for the ventilator. But I can shed some light on various “why do they do that”.
JPL doesn’t do manufacturing – in fact, they are prohibited from doing so, because we’re prohibited from competing with industry. Besides, it’s not our thing: JPL does bespoke artisan hand crafted spacecraft – typically as one-offs – we don’t have people who do production line stuff – JPL does do some board fab and assembly for prototypes, just like you might in your garage (just with bunny suits and strict process controls), but most gets contracted out to local companies who do it all the time.
As for software production practices – sure, JPL flight software is produced under fairly rigorous processes, but JPL, as a institution, produces an enormous amount of non-flight software, which is NOT produced under rigorous processes. One hopes that “good habits” carry over, but to be honest, the kinds of things one cares about for flight software are similar, but not the same as a medical device. Medical device software isn’t likely to see single event upsets, for instance. Spacecraft software makes heavy use of watchdog timers of various kinds, but for the most part, operations in space are slow and stately (except for entry, descent, and landing). If the rover software has a hiccup, and has to pause for 5 minutes to reboot and reset, that’s probably not a crisis. If a ventilator stops for 5 minutes, you’ve got a injured patient on your hands. That said, JPL does produce timing and function critical software, and does it well – but there’s a lot of software, even in flight, that has different tolerance to unusual events.
Everything JPL invents/creates is covered by the Bayh-Dole Act, because Caltech is a non-profit educational institution. All government funded research is subject to the Act. That means that Caltech has the ability to license IP developed at JPL – the government gets a “fully paid, royalty-free, non-exclusive” license, but Caltech can license it to others. The government also has what are called “march-in” rights – they can take the IP and give it to someone else to use for government purposes. There’s whole books on how narrowly or broad “government purposes” might be.
Typically, the license fees are quite small – Caltech isn’t looking to build their endowment with this – more, what they try to do is get the technology into the market place – and sometimes the best way is to pick a few companies to license it – they get exclusive licenses, so they can invest the substantial money needed to set up production, training, and support operations. There is almost always a clause in the license that requires that the product be “on the market” and being sold within a certain time. That’s to prevent a company from doing a “license it and put it on the shelf” strategies.
As far as whether the JPL ventilator, or anything else JPL does, might infringe a patent, I can’t speak to it. It probably depends on what the government is doing with the IP of others. I would expect (but do not know) that Caltech will NOT indemnify anyone who licenses the design, though. That’s one of those costs of “setting up manufacturing” – making sure you’re not stepping on someone’s proprietary rights. As we all know, there’s no exception in the patent or copyright laws that says “you cannot use this without permission, except in situation X”. However, I would imagine that under the authority of the Defense Production Act, or something like that, there might be some provision for mandatory licensing – the patent holder would get paid, but couldn’t prevent production. Sort of like a compulsory license for copyrighted songs. I’m not a IP lawyer, and this is really complex stuff.
Hi, thanks for this great act. Could you help me how can we get the license for production from JPL?
The stats for people that are intubated making it are not very good. Only a fraction. The virus is now looking like it just uses lungs to get in to the body and then wreak havoc in the epithelial cells. That includes the interior of the blood vessels. That’s why there is significant percentage of the patients having clotting and strokes, cardiac issues, etc. Usefulness of the ventilators are not as high as we thought.
https://youtu.be/Aj2vB_VITXQ
I can imagine that conversation… Come on guys, let’s design a ventilator, it’s not rocket science.
I’m part of a team working out of the University of Florida that is seeking to build a low-cost ventilator for Third World countries where resources are scarce. Many of us are ham radio people with EE or software backgrounds. (I’m part of the software team.) You can read a little more at: http://www.arrl.org/news/radio-amateurs-team-up-to-help-university-design-low-cost-ventilator
Our unit is ready for FDA submission and we think we can build the units for $200 or less. Ours doesn’t look as pretty, but the unit has been stress-tested, in some cases to more than 2.5 million cycles, without failure. You can get more info by searching “university of Florida ventilator project”
Dr. Jack Purdum, W8TEE
Paid no-doubt with public money, yet not open hardware. Shameful.
https://techcrunch.com/2020/05/01/nvidias-top-scientist-develops-open-source-ventilator-that-can-be-built-with-400-in-readily-available-parts/amp/
Also see “Mechanical Ventilator Milano.”
https://arxiv.org/abs/2003.10405
Any estimates for manufacturing cost?
SARS-CoV-2 infects and damages literally every single major organ. heart, lungs, kidney, blood vessels, brain, immune system, even testicles. That said, if you need oxygen, you need oxygen. Patients won’t get a better outcome because they didn’t have access to a ventilator when they needed one.
But the best solution is still prevention. An ounce of prevention is worth a pound of cure. Wear a face mask to protect yourself and others. 60 to 70% total infection rate is civilization ending, don’t let it reach that point.
Good luck with prevention unless a vaccine comes soon (months) and is very effective. We should expect a close to 100% infection rate in the end, hopefully with a weakened version. Calling it civilization ending is hyperbole not supported by public data and a face mask offer little protection.
With such a short Time to approval I am fairly certain there is no software involved at all. This also justifies the use of segment displays. This also explains the short life span as it is likely assessed as non PE device. Four months is probably the calibration interval and hence as it is a non programmable device it can not be recalibrated and must be decomissioned.