Even theoretically, you would then need to have perfect information every single other factor influencing relevant-demographics death rates, assuming you somehow magically know the exact relevant demographics. If there is even one other factor that is uncertain, you end up having to increase your approximation's margin of error proportionally to the uncertainty, and each missing data point is another power factor of increase in the margin. Eventually, it's much smarter to realize that you don't have a clue.

Now, take into account that you don't even know all of the factors, and that it's pretty much impossible to prove that you know all of the factors even if by some unknown feat you managed to figure out all possible factors... quickly the problem becomes far beyond what can be calculated with our puny mathematics, let alone by a human. Of course, if you still just want an approximation after doing all of that, it may become possible to obtain an accurate one, but I'm not even sure of that.

Thanks for the food for thought, though.