There's a lot of junk "science" used in trials because there are plenty of "experts"[0] available to back it up for the prosecution and fewer funds to pay for the countervailing defense expert available to present the problems with it. Usually it takes a particularly bad case making it to a supreme or appeals court for that kind of evidence to be disallowed.

[0] The people paid to perform these analyses in the first place and then go testify convincingly for the prosecution about it but that's a whole separate rant.