Great post! I really like the care you put into exploring the data.

One aspect I would be careful about is saying the accuracy numbers are good. When we are talking about the prediction of death, say to decide how to triage care, that would be wrong for way more people than would be acceptable.