Some notes/highlights (but it's not that long and you should read the whole thing)*:

• Generally thought to be more helpful if you can get both oxygen and pressure (pressure helps keep alveoli open)
• At least several oxygen supplementation methods will aerosolize the virus and increase contagiousness nearby, and are not allowed in a healthcare setting as a result. Perhaps particularly methods involving pressure.
  • This is not, or is less of a problem, for low-pressure oxygen (ex: (non-rebreather) oxygen masks). These oxygen masks are probably what is meant by standard oxygen therapy.
  • This is a problem for CPAP machines
  • (Deleted a somewhat-incorrect/confused bullet-point)
  • (Personal thought: Is this part of why hospital outbreaks were so nasty with SARS-1?)

suppose you’re a person with respiratory failure, but you have an oxygen mask, CPAP, BIPAP, or high-flow nasal cannula at home. 40% of the time or so, your noninvasive home equipment is going to fail and you’ll need to be intubated. Usually this happens quite urgently; if you need to take an ambulance to get to the hospital it might be too late.

• Invasive intubation tends to be be urgent, and is a very seek-a-medical-professional do-not-do-this-at-home procedure
• From ~300 pneumonia cases: From the set [oxygen mask (rebreather), high-flow nasal cannula, non-invasive ventillator], high-flow nasal cannulas have the best outcomes associated with them (much better 90-day mortality rates). But for all of them, 30-40% ended up needing invasive intubation within 4 days.
• Given the frequency with which people go on to need invasive intubation, hospitals are probably better than at-home with NIV unless you really can't access the hospital (ex: overun). The rates for needing intubation are still high (~30%-40%).
  • But see also: NIV seems better than standard oxygen therapy (oxygen mask), in the follow-up post

Many of these [open-source ventillators] are automated bag masks, or CPAPs, which aren’t recommended for use in hospitals because they spread COVID-19; and they won’t work on the more severe cases of respiratory distress anyway. Only two projects that I found, the Pandemic Ventilator and the Flometrics project, are explicitly trying to match the specifications of the type of mechanical ventilators found in ICUs.

The follow-up post on non-invasive ventillation (NIV)

• Lot more data surveyed, but across several ventillators and conditions (ex: COPD (definitely helps), pulmonary edema (somewhat less consistently)).
• NIV does seem to improve outcomes by quite a lot, including in pneumonia.
  • Strong results on intubation there (60% changed to 15-20%), but only 136 total pneumonia samples across 2 studies. And one of them noted no difference in mortality.
• NIV makes a pretty big difference on the overall intubation rate, changing 30% to 15%

This means that, in a world where ICUs are overrun, there’s a good chance that a home non-invasive ventilation device could save lives for people with COVID-19.

* Also, I'm not likely to sync this as it updates, I may be out-of-date, errors in bullet-points are mine, etc. etc.