Using machine learning to detect deficient coverage in colonoscopy screenings

This really resonates with where I feel a lot of these systems are headed first. Having a system that aids the reader/user or augments their own abilities are far more palatable to a system with tons of regulation. Imagine a chest xray scanner that highlights places that are typically missed by radiologists, or enhances the opacities in that area. If I'm not mistaken, philips or siemens have a program that does just that.

Late 30s. He'd been battling it for four years, and it was discovered at Stage 3. Which would suggest it'd been growing since when, 37 or so?

Indeed. In Chadwick Boseman’s case, the onset was at _such_ a young age that it suggests (total speculation) he might have had a hereditary cancer syndrome or some other familial risk factor. More widespread genetic testing may help people like him in the future.

I wonder if a better starting point would be consulting a well-regarded oncologist or gastroenterologist.

That might be your best resource for assessing your personal risk profile, and developing an early-detection strategy. That also might be the fastest way to persuade a gastroenterologist to perform the exam, if appropriate.