2 options:
You could treat it as a 3D game with a fixed orthographic camera. Then you could use any standard 3D games resources / books / tutorials / libraries. Even if it rendered in 2D with sprites, the maths for movement etc would still be 3D behind the scene. This would give you the most accurate results.
You could treat it as a top-down 2D game, with an added height property. For example, imagine a 2D grid where each tile has a height value. When moving from one tile to another, you only let a character pass to a particular tile if the characterscharacter's feet are above ofor equal to the height of that tile, but if they are jumping and gravity brings them down on top of a tile, you let them remain there. With a bit more work you could make it so you can walk under bridges etc.