I wasn't necessarily assuming that Perseverance would be functional to hand the samples over; rather, I thought it might intentionally store the onboard samples in a fashion where another rover could easily grab them without direct intervention from Perseverance itself. Something like a tray or rack of tubes with each sample having exposed grips.

The comments made by others suggest that Perseverance does have onboard samples like this, and the dropped ones are secondary (in case something happens to Perseverance to destroy all its onboard samples -- I don't know, catastrophic damage to its radioisotope power generator? Or perhaps more realistically, it gets stuck in a sandy area which is too dangerous for other rovers to approach without also getting stuck), which makes more sense.

Maybe someone can explain to me: why does Perseverance drop samples, instead of keeping them stored onboard? Wouldn't it be easier for some future rover to just go to wherever Perseverance is and collect all the samples at once, rather than needing to trace Perseverance's entire journey to pick up the samples?

Accuracy's isn't what's important, precision and recall are. Since to get used, the model needs an appropriate threshold, and then you must consider the consequences of that threshold.

Forget issues of using socioeconomic data, which bakes in existing biases, the top 10 features in their model are variations on "past misuse of drugs" or "long-term use of opioids." And #11 is Chronic Pain. So is this actually telling clinicians anything they don't already know?