I don't think a breast cancer screening study would cost hundreds of millions.

That's the wrong analogy; we're talking separately selling a device that tells you some metric, not a diagnosis like breast cancer. So presumably you have to compare it against some existing way to measure that metric.

I don't know how much testing & FDA approval normally costs for nontrivial "metric only" devices (e.g. glucometers), but on the outside view, I doubt it's cheap as you say, because everyone in the system has incentives to raise the price, just like for drug trials.

Especially at the beginning you would have most of the software being developed by the company who sells the device. Business wise that means it might be very worthwhile for a company who develops this technology because they might have a monopoly on the marketplace.

And product wise that would result in the same kinds of products we have now, i.e. extremely conservative and slow moving. And approved and marketed for just a few uses, instead of being open-ended like you suggested.