People also die because of late or never materialized medical technology. We just prefer that people die from inaction (because is "normal") than from mistakes. This may be psychologically attractive because we avoid SEEING all the blood and dead bodies, but probably not the best way to minimize death and suffering.

Of course this is nothing new (Economic version: "That Which is Seen, and That Which is Not Seen"). Just another cognitive bias.

There is some more-or-less optimal middle ground between too little and too much regulation, but I don't think we're on it now. In fact, maybe it would be worthwhile to have a two-tier system, one for ordinary medical devices, another for experimental "this will probably kill you but perhaps you have nothing to lose" situations.

The most scary moment - one of the applications we've written was used to take life-saving medications on time. Scary stuff I'm telling you.