I also doubt the real gain is in spectacular cases such as the brain tumor in this example.
I find it way more likely that we will see some advanced version of IBM's Watson use that data to cross-correlate data and suggest things like "with your genes and current condition, you should use antibiotic X, rather than Y, even though Y is the better choice for most." or even "you should stay home today, or you'll be ill for 2-3 days."
Each such choice may only give a small gain, but there will be way more choices to make.