Air Collective Forum

Could a doctor please review this, to inform an engineer how to improve their respirator prototype?

These questions come from Johnny Lee, who’s created a model based on converting a CPAP blower. (click the link for more info on his device)

He would like a separate medical opinion to correct or improve his understanding as needed, and replies here will be open to inform all other engineers working on respiration devices.


I had a long conversation with a pulmonologist to discuss the critical features of a proper ventilator and trade offs. … If you know someone with domain expertise, I would like confirmation that my take away notes are reasonable since this will drive priorities on engineering effort/conversations.

On 2020/03/18, I was able to spend a couple of hours talking to a pulmonologist to better understand the critical functions, and concerns in operating a full featured ventilator. Here are my summary notes and my non-expert interpretation of the discussion. This is not direct medical advice. I welcome additional expert opinions.

1. Prioritizing professional care is ALWAYS advised.

If professional care is available, you SHOULD NOT pursue using anything like this. This is not an alternative and should not delay seeking available care. You put yourself, the patient, and other caregivers at risk. Projects like run the risk of making it too easy to be harmful. I fully agree with this. I also really appreciate medical professionals willing to engage in the discussion/education of folks like me (and interested parties) about what the risks are, what the trade offs are, how important are different factors, and how they could be mitigated with more educated usage or better technology. Thank you to those who are willing to engage in the education.

2. If configured reasonably, a device like this could be helpful if you are “up a creek without a paddle” and professional care is unavailable, in self-quarantine conditions, in early onset, or in the recovery stages when critical support from a real ventilator is no longer needed.

Proper ventilators have many more features that allow physicians to specify things like target absolute volume and pressure cycles tuned to the specific patient, O2 percentages, tools to monitor the compliance of the lung tissue, as well as balancing other risk factors that might cause barotrauma. This includes considerations around non-uniform tissue strength in the lungs which might rupture, presence of liquids, pressure spikes due to coughing, mechanical issues in the chest cavity such as the rib cage, and a host of other conditional complications that might require exceptions to any rule. As you go beyond 20cm/H20, the risk that patient specific issues and lung health conditions could result in tissue damage goes up significantly. Again, prioritizing professional care is always advised

At 40cmH20, over expansion risks with traditional ventilators are high enough that treatment should consider moving to High-Frequency Oscillatory Ventilation (HFOV) which replaces large inhale/exhale cycles with rapid (3Hz-15Hz) small breaths at medium pressure. This eliminates the extreme expansion and contraction of a typical ventilator minimizing mechanical stress on the lung tissue. So, every piece of equipment has its place and its limits.

In the event that ventilators are not available, and other hospital equipment that could act as a ventilator (such as anesthesia machines) are exhausted, non-invasive ventilation such as CPAP/BiPAP machines do have a niche role in helping people that are not in need of high pressure (30cmH20) care. These might include early on-set stages of illness, recovery stages after high-pressure care is no longer necessary, or in environments where there are no alternatives such as self-quarantine. Additional sensing is required to make it less likely to cause tissue damage at pressures above 20cmH20.

Note: I’ve read anecdotal evidence that BiPAPs were helpful in Wuhan, but it was not authoritative evidence to reference.

3. At a high level, the goal is to get enough air into the lungs to keep the patient alive without causing unnecessary stress on infected lung tissue.

Pushing air into the lungs is an attempt to force the lung alveoli into continuing enough O2/CO2 exchange to keep blood O2 levels at an acceptable level. However when the lungs are infected, the tissue compliance is reduced requiring more pressure to reach the same volume of air. That increase in pressure also increases the risk of damage. It’s like a balloon with rubber material that is getting stiffer and more “brittle” which can lead to “cracks” causing it to pop when you blow too hard. So you have to strike a balance between getting enough air into lungs to survive without pushing too hard damaging the tissue. To monitor this compliance, ventilators have something called a loop display [from Fundamentals of Mechanical Ventilation, R.Chatburn]

The left chart shows the loop created in a volume-pressure plot created by the inspiration/expiration (i.e. inhale/exhale) cycle of a breath. Like a balloon, during inspiration, it takes a bit of pressure to begin the expansion: then as the lungs reach capacity the volume increase slows relative to the increase in pressure creating an S curve. On expiration, pressure drops rapidly as the lungs contract and returns back to its original state. This relationship between the pressure required to achieve a specific volume, particularly in the inspiration phase, is a function of the compliance of the lung tissue. As the patient gets less healthy, the lung tissue compliance decreases, requiring more pressure to achieve the same volume:

The blue cycle is a healthy individual. The red cycle is an unhealthy individual. As you can see in the unhealthy condition, increasing pressure has diminishing returns on achieving more volume. In the blue/healthy cycle, the target volume is reached with much less pressure and never flattens out. The lung is able to reach target volume easily. This plot is one of the main ways physicians can determine “how much is more pressure worth it?” and “how complaint/healthy is the lung tissue?”. The slopes of these curves are critical feedback to determine what is an appropriate volume and pressure ventilator setting. It also changes over time as the patient’s condition evolves.

As shown, these loop displays require absolute measurements of both volume and pressure which are difficult in a BiPAP setup where there is leakage at the mask/mouth piece. However, the goal is to get an indicator of lung compliance which is actually the ratio of L/cmH20. This ratio is not an absolute measurement and may achievable with clever low-cost sensing techniques comparing relative measurements avoiding costly calibration. TODO: think about how a reasonable proxy could be achieved cheaply/easily. I could use help with ideas here.

Initial setup of a ventilator often starts with testing an educated guess given the properties of a patient, watching how the patient’s body responds - such as observing the chest rising or watching where the loop display curves flatten out. Then the settings are then tweaked to “give it a little more” or “back it off a bit”. A simplified interface to bump the pressure up/down with feedback on lung compliance might be sufficient in an emergency situation.

4. Alarms, Alarms, Alarms. Alarms are important.

Highest priority: If the mask falls off/gets dislodged, sound an alarm. Next highest priority: If the target pressure/volume/ratio is not getting reached, sound an alarm. Nice to haves: if the patient is coughing excessively, sound and alarm. If the patient’s lung compliance is changing significantly, sound an alarm. Notifying either the patient or a caregiver that “something is wrong” is a high value feature.

5. At lower pressures (~25cmH2O and below), intubation typically has more to do with the safety of others rather than a necessity for treatment to the patient. When aerosolized virus is not a concern, such as isolation or an already “dirty” environment, masks and mouth pieces are likely okay.

Non-invasive interfaces like masks and mouth pieces are likely to aerosolize droplets containing the virus upon expiration (exhalation) spreading the infection to others in a clean environment like a hospital. Intubation captures the expelled air and filters it, preventing the spread of the infection. At higher pressures, there are mechanical reasons why intubation is advantageous. But at lower pressures, it is not meaningfully advantageous to the quality of the treatment to the patient. In fact, the invasive nature of intubation exposes the patient to additional health risks. If professional hospital care is available, that is always preferred.

If the patient is in isolation in a single unit home or if the surrounding environment is already “dirty” with many other sources of aerosolized virus (like an open tent of coughing people), a non-invasive interface like a mask or mouth piece is likely less problematic. Staying at home in a apartment is NOT true isolation due to potential shared ventilation with other units. Capturing the exhaled air and filtering it is preferred and may make it palatable to use in more environments. Since this device is not going to be FDA approved, it is not likely to be used in a hospital anyway.

6. Triggering the BiPAP on patient inhalation is more about patient comfort and may not be necessary to be effective. If the choice is between being uncomfortable or not being able to breathe, people will likely choose being uncomfortable.

When the BiPAP inspiration/expiration cycles do not align with the breathing cycle, it is uncomfortable because the machine “fights” with the patient. Modern BiPAPs can detect the beginning of a breath and let the patient drive the breathing cycle enhancing comfort. But based on the conversation so far, this did not seem to be essential for useful treatment. A conscious patient can synchronize their breathing to the machine. It is not clear what happens when the patient is unconscious, such as sleeping. The current hypothesis is that it may lead to a poor nights sleep but the patient will still receive airflow as long as the set rate is reasonable. Ventilators have a “backup rate” setting which forces air into the patient in the event they don’t take a breath within a target time period. This is the same as the open loop BiPAP operation.

7. The only difference between BiPAP and PEEP is that BiPAP is the terminology used with a non-invasive interface (like a mask) and PEEP is the terminology used with intubation. Otherwise they both refer to having a set positive pressure during the exhalation phase.

8. Open Source Medical Device Guide

A Facebook group was setup to discuss recommendations around open-source efforts in response to COVID-19. Here is a 1.0 version of the doc they released. Most of the guidance so far is written in a way that assumes professional care is available.

(more to come)