It's called the longitudinal study. More precisely it's a specific type of longitudinal study called a retrospective study.
The method is simple. Take a big collection of data and test it for some other purpose.
Here is one from today's news. Old people don't need the flu shot. Someone ran the data and found it doesn't help you live any longer.
Maybe you can throw a few new assumptions into the data, as with my recent piece on the HPV vaccine. Or you correlate disparate bits of data, to find out which immigrants are depressed. Or you want to turn some assumption on its head.
Such studies need more than mere compute power. They need databases to run against. Big studies, with lots of people, and extensive documentation on each patient.
Right now all this makes the data from old studies highly valuable. But the advance of EMRs (or Electronic Health Records, if you prefer) is going to change that.
EMRs can be collated into huge databases with which, through programming, large conclusions can be drawn, without ever violating the privacy of anyone. Or even seeing anyone.
It's the sort of work politicians have been doing for decades, only with much greater depth, because the database is deeper, and with a purely scientific motivation.
Whole new fields of study, like predictive health, are emerging from this process. But we're talking here of a very depersonalized study, one that doesn't resort to actually seeing patients or diagnosing their ailments.
It's the kind of work computers were made for. And it's only going to become a bigger story as more databases are built and more questions are analyzed. We're on the threshold of a new kind of science here, and a new set of scientific dreams.
You really are magnetic ink.