(05-22-2021, 11:48 PM)Crimson Nape Wrote: During the early days of Covid-19, it was reported the Resmed firmware is what differentiates the different models. The mechanical components are the same. This came out when there were discussions about converting a CPAP to ventilator.
Wow -- I had forgotten this -- but yes, I remember seeing those reports, too!
You know, this brings up something that has bothered me since I first found out that the manufacturers were producing data-logging and non-data-logging versions of what they
claim are machines with the same algorithms.
Ok, I've been a professional software developer for 36 years now. When I look at the information on that SD card in my machine, I'm seeing something that I recognize as a fundamental tool of my profession: print statements in the software writing out all of the data that the machine is using, and logging what it is doing as it runs. The purpose of these print statements is that this is -- practically speaking -- the ONLY way to know whether the software actually does, in real life, what the programmers INTEND for the software to do in reaction to all of the real-life ways that an apnea patient's body can behave while asleep.
When I look at the A10 Auto vs the A10 Elite, the only way that I can
imagine that these two products could
possibly have been engineered is this: First the engineers programmed the Auto prototype. The engineers and the QA team spent thousands of hours testing the software by pouring over the data on the card, testing every code path, and making sure that the machine did what it was supposed to do. Over that process, hundreds, thousands, tens of thousands of bugs were discovered, diagnosed, fixed. Then when the machine PASSED this QA process, they built a production line that could produce the prototype in production quantities. AFTER the SD card and its data had served as one of the bedrock parts of the QA process the engineers took their prototype and got their soldering gun out and unsoldered the connections that connected the SD card drive from the circuit board on the Auto, and it became an Elite.
So now you have a machine where it is impossible for the QA team to do any process that uses the card data to double-check that the machine does what it's supposed to do.
They just
assume that ripping out the SD card drive does not cause latent bugs to manifest themselves.
The FDA just
assumes that the Elite (the Elite that is impossible to test) works just like the Auto that has been tested.
The doctors, patients, insurance companies, DMEs, etc -- everyone has to just
assume that the Elite does what the Auto does, even though there is no card to check.
...and we all know that when you
assume you make an
ass out of
u you and
me!
One of the most fundamental rules of safety engineering is that
any change to a design, no matter how seemingly trivial, introduces the possibility of introducing a new error, and so requires a new, separate QA process post-change. Safety engineering is all about making it a pain in the ass to change anything, so that
nothing gets changed without a damn good reason!
Ok, now we are talking about a
medical device which is quite complex and sophisticated -- unlike something like a scalpel which is fundamentally just a sharp blade with a handle. It is used by completely untrained people outside medical observation -- unlike something like a robotic surgery device where the surgeon has years of training and experience. Its normal operating environment is a room where
no one is conscious -- so even an "obvious to the most casual observer" problem will not be detected because no one is watching.
As a software developer who has spent literally thousands of hours tracking down software bugs, I will say that there is
no fricking way that I would fall asleep using a cpap machine that does not produce detailed data. Ok, maybe if you could prove to me that the Elite is an Auto where they super-glued shut the door to the SD card slot, ok that I would accept as a trivial enough change that I would trust the machine.
But even that brings up a whole different problem: when the "brick" version of a machine is simply a somewhat-crippled copy of the full-function machine, the process of crippling the "brick" means that the "brick" is
more expensive to produce than the full-function machine. (The savings of not buying the SD slot hardware is a few dollars. The engineering cost of doing this properly is orders of magnitude more than that.) So either the ResMed people are fricking insane, or the safety of the Elite is in serious doubt. Either way, I'm not going to sleep! (And, actually, the prospect that the ResMed people are fricking insane gives me serious qualms about strapping on a mask hooked to
any machine that they have engineered.)
...wow, that turned into quite the middle of the night rant!
