Right I used CPAP as an example because it bypasses all arguments about "novel technology", "drug development" cost, or "need for safety". Even an ASV algorithm could probably be implemented as a ~graduate project since it seems to be a rather basic control algorithm (today with ML you could probably do even fancier things). It's basically a piece of pure technology that could be commoditized, manufactured and sold to people for < $300 in the way smartphones are. If something as "basic" as this can't be accomplished due to the regulatory environment then it's sort of pointless talking about anything more complex like pharma drugs.