You can't have any impact at all if you can't legally operate it or if you are hamstrung in your marketing claims.

Chatgpt is just words. This is an ultrasound imaging system. Who knows what could go wrong: blown out eardrums from feedback, acoustic burns, wild inaccuracies that lead to misdiagnosis.

There's really no way around documentation as a way of collecting evidence that the team knows what they are doing. Things like enumerating all the possible patient risks, assessing their severity, updating the design to mitigate, and ultimately testing that it works as intended.

This is why you can't just bolt on the medical device part. Most devs will have a conniption if suddenly expected to attend lots of meetings and do a lot of paperwork. Different skillset and very expensive to switch out your whole workforce.

Let's say you do a diagnostic based on the results of this machine. Later, patient suits and you have to answer for your diagnostic. If the only evidence for your diagnostic is coming from a non-FDA approved machine, you're liable.

It needs approval from someone to ensure it's safe first and foremost - health & safety inspections, hygiene, FCC approval, etc.

And FDA approval (I presume) if they want to give formal diagnoses, but I believe that if they don't get that it'll fall under the "alternative medicine" umbrella, which is very broad. But they can do whatever under that umbrella as long as it's safe.

I wonder how doctors feel about people consulting ChatGPT. I know a doctor who can't stand people getting medical advice from quacks on TikTok, but that's probably a different problem/pattern