A Quick Comparison of the AmboVent and Apollo BVM pandemic Ventilators
Last week, the first week in April, 2020, two projects, AmboVent and Apollo BVM released open-source designs for last-resort pandemic ventilators. They (and other fully open teams) deserve our profound thanks. At EndcoronaVirus.org and Public Invention, we have been maintaining a spreadsheet which seeks to give some visibility into the many open source projects underway right now. Both of these projects are fully documented and highly reproducible, and can be easily manufactured. Policy makers, NGOs, and governmental officials may want a quick comparison beyond the evaluation we provide in our spreadsheet.
Open-source ventilator designers will want to study these projects, and, I hope, improve them. The beauty of open designs is one team can easily improve on the work of another. I encourage all teams to release whatever they have as soon as possible for the good engineering community and people afflicted by COVID-19.
With some humility and no authority, I present the following observations.
My executive summary: The AmboVent is closer to being clinically suitable and ready to be manufactured by a government or NGO in the 1000s. The Apollo BVM is easy to reproduce on a small scale anywhere. The AmboVent has better testing and better alarms and monitoring at the time of this writing. The AmboVent would be easier to use by exhausted front-line clinicians wearing PPE. The AmboVent is closer to meeting the important RMVS spec, by a hair.
The AmboVent was made by a team of about 40 professional engineers, makers, doctors and innovators in Israel. The Apollo BVM was made by senior undergraduates assisted by faculty and staff and Rice University and Rice Oshman Engineering Design Kitchen. It was inspired by a similar project at Rice from 2019. (Full disclosure: I graduated from Rice in 1987. Go owls!)
Both ventilators are “Bag Mask” or AmbuBag designs. That is, they use a common, off-the-shelf device called a bag mask or AmbuBag. It is a self inflating bag with two valves and a mask. You fit the mask over a patient’s mouth and nose and squeeze it with your hands to force some air, often with oxygen or 100% oxygen, into the patient’s lungs. This is normally done in traumatic situations away from hospitals until transport to a hospital, where different devices are used. Both automate the process of squeezing the bag. This design has strengths and weaknesses.
The basic design has the advantage that the volume of a breath can be easily limited. If you blow too much air into the lungs, especially of a patient with ARDS, you damage them. It has the weakness that we do not know how long you can squeeze an AmbuBag before it loses its natural ability to return to its shape. A computerized pump might fail due to a software error; these designs do not have that weakness, though they may have other problems.
The Ambovent uses a lever arm to press downward on the bag. The Apollo BVM uses two rack and pinions to press the bag from both sides. This means that the airway hoses don’t wiggle, which is good, but it has more moving parts, which is bad. I believe the rack and pinion may be a better way to squeeze the bag, for what that this worth.
Most non-pneumatic ventilators designs have a microcontroller. Both of these use the ubiquitous, open source hardware Arduino family. The Apollo BVM uses two Arduino Unos, available all over the world in high quantities and scroungable from maker spaces. The AmboVent uses the Arduino Nano. The AmboVent mounts the Nano on a printed circuit board that is simple; it could be etched in a maker space or mass-produced with normal PCB production. I’ve been told you could wire it by hand easily if you had two. The Apollo BVM uses two Arduinos and at this writing connects them with jumper wires to a breadboard. This is great because it can be easily duplicated by almost anyone with any experience. That approach is too fragile for clinical use, IMHO.
The AmboVent is controlled by knobs you turn. It is similar to older ventilators that clinicians will be familiar with. It is a robust and simple system. The ApolloBVM uses tiny buttons that would be hard to use when stressed and wearing gloves. The AmboVent display seems very small. I would like to see a standard user interface emerge, that is used by all teams or most teams to minimize training effort. I encourage the Rice team and other teams to copy the AmboVent design in this regard.
Alarms and Monitoring
The AmboVent has them; at this writing I see no evidence that the Apollo BVM does. I believe the AmboVent has an internal airway pressure sensor, and I don’t think the ApolloBVM does at present. Public Invention has created an early-stage open project, VentMon, to modularize alarms and monitoring so that teams can reusing similar monitoring solutions. Alarms and monitoring are integral to testing and clinical use and absolutely mandatory according to the RMVS spec.
The Apollo BVM uses laser-cut plywood which will be easily made in small units. (As I was writing this, Amy Kavalewitz showed me a photo of an acrylic case — these teams are moving fast!)
The AmboVent uses formed aluminum sheet metal, which is also easily made. The AmboVent’s case is better for a hospital environment in terms of disinfection.
I believe the Rice team is doing pressure/volume testing as I type; the Israeli team has done a great job of this already. Testing is essential in last-resort ventilators because they are life-critical equipment in which a bug or mechanical failure can mean a fatality. Likewise, not all AmbuBags are the same. Supply chain disruption means open-source pandemic ventilators have to be more robust and accept more variation in their components than any modern manufacturer ever had to consider. There is an urgent need right now for an independent team to build and test the AmboVent to independently verify its performance, the clarity of its documentation, and its reliability.
Both systems fail to meet the RMVS without additional equipment, because they cannot take pressurized air from a clinic wall at 50 psi (American Standard) or 4.5 bar (65 psi) (British standard). They each would require a free-standing air-oxygen mixer of some kind. The recently released Medtronic ventilator also fails to meet this spec and does not support 100% oxygen.
I believe the AmboVent meets most of the RMVS spec. The Apollo BVM needs monitoring and alarms to be able to do so.
In order to be reproducible and testable by a third party and to help the community, a design must be well-documented. This is a primary concern, but I have saved it because both teams have done a great job!
Gratitude for the Teams
Everyone in the world is doing their part; engineers and makers are doing their part. I would especially like to thank the Rice and Israeli teams for making their designs fully open so that we can learn from them. If I have said something inaccurate or that soon-will-be-wrong I apologize.