I think if you read this article you'll see (at least part of) the answer. Not only does a given action have to cause a definable result, the medium upon which the result is deposited has to provide enough detail to read the result back off it, and the people reading the result have to be able to be consistent and accurate in their analysis of that result.

With bite marks, the skin has to hold the bite mark (which it doesn't) and experts have to be able to map it back to real human teeth (which they can't). With blood spatter, there are analogous problems.