Your thoughtful question is definitely along the lines where the research could change health outcomes. Apple Health currently tracks trends over time and can alert if any disconcerting trends are identified. If Apple were able to capture a breathing signature at rest, say once a month, trends could be identified (via training data, as you mention) and data optionally provided to healthcare providers.

Some people who are alone (including my father) have no idea that they have sleep apnea or 'odd' breathing - for apnea, they're obviously asleep, and for other breathing factors, it's usually a slow and unnoticeable progression.

> is there any sensor data from stuff like apple health care that could be put into an ML to detect such changes on breathing fingerprint?

Pixel (all of android?) devices have some sleep-time snoring/breath interruption detection built in.

As far as I know, they do this with just microphone so the apple watch should be able to do this, too.

Plotting the measurements over time would be really cool!