One point I can make is that you'd have to have many experts from different fields to test a variety of different ideas. They couldn't guarantee that the idea will work in practice 100% or even that it won't work, but if you came to the expert of a certain field, who has a broad and deep enough knowledge and practice, with an idea directly related to that field, I assume he/she would be able to tell you pretty well what your idea is missing, which main steps you should take to approach realizing it and what are the approximate odds for its success.
Get a second and third opinion from different, unrelated experts of the same field to paint a general picture.
So connect the experts willing to participate in such idea validation business through a dedicated network (online or otherwise), pay them enough, build an infrastructure to conveniently consult customers with their ideas, and there you have it.
As J. Nikola noted, you can't be sure unless you test the idea in practice, but that would perhaps be as close as you can get before testing in practice.