Better reliability of "Adduct/Charge" predictor

Currently, during nontarget Analysis, Analytics predicts the charge state of an unknown based on other peaks in the isotope cluster or charge envelope. When the software finds the monoisotopic peak, it does a fairly accurate (90-95%) prediction of the charge state (i.e., M+H+ or M+H2+). However, the software commonly chooses the second or third isotope peaks in a multiply charged cluster as a unique compound and predicts the mass to be M+H+. This causes the nontarget result list to contain many false positives that must be individually screened. Similarly, peaks with the charge state M+ are always misidentified as M+H. Mathematically, this would be very difficult to predict the difference, but maybe there is a way to include this in the algorithm, or manually adjust it if the compound classes are commonly M+.