There are a lot of mathematical models trying to approximately predict the water activity levels ... one of them is Grover's model, very simple and in my calculations it gives really good approximations. I cursorily tried it with the recipes from the J. P. Wybauw's "Great Ganache Experiences" book which do not include fruit purees. For some recipes, it really gives good predictions ... but for some not so good. I think that the differences mainly come from the constant used in the model for glicose syrup ... and for some recipes, I also think that there may be some measurement errors for aW values given in the book.
The constant for sorbitol is given as 1.3 in some other sources ... in my opinion, this constant should as well be taken with a grain of salt.