"Useful" means different things in science and in patent law. As one court put it, "anything under the sun made by man" is useful in the context of patents. Accordingly, much of the arguing is over other requirements, namely novelty and non-obviousness.
It's true that research papers contain a lot of data "proving" usefulness. But research papers have a different purpose than patents do.
As JimIvey points out, "useful" in the patent context has a very specific meaning. "Usefulness" is a problem for very few patent applications -- mostly ones claiming oddball stuff like cold fusion and perpetual motion.
The big issues which plague inventors in their efforts to obtaining a patent are novelty and non-obviousness. Also, to a lesser degree, enablement (how to make and use the invention) and written description (loosely, what did you invent?).
The market will decide whether or not things covered by your patent are useful.