GREP can't find things like that - it can only find at the beginning of a word/paragraph or at the end of a word or paragraph - but it doesn't know when something is approaching the end of a line within a story.
It will either find a string of numbers or nothing at all - it can't descriminate between, 1 apple, 100 mgs and 1000 mgs and 2004 (year).
It can find just a Number (\d) or a string of numbers (\d+) or you can instruct for (\d{3}) to find 1 or 3 digits etc.
The next piece of the puzzle is the space which is \s+ for a constant string of white space, which also includes tabs and other space characters. If you want just a Space then you have press the space bar so it looks like this (without the brackets) (_+) (the underline represents a blank space by pressing the space bar).
So the GREP solution would have to find any number followed by a space and either replace the space with a non-breaking space or a Apply a No Break in the style.
The problem with the Script Solution and the Find/Change is that I imagine you'd have to it multiple times and if you forget to do it once then you end up with a number being left behind on the line above.
I'd still be inclined to use a GREP style to change the Number followed by a Space into a Non-Breaking character style.
