A method to diagnosis a disease state of an unknown sample. A test Raman data set for an unknown sample is generated. A reference Raman database is provided where the database contains a plurality of reference Raman data sets and a plurality of reference Raman difference data sets. The reference Raman difference data set is generated by determining a difference between a first reference Raman data set and a second reference Raman data set. A first reference Raman data set is associated with first known sample and associated with one or more of: a first known disease state and a first known clinical outcome. A second reference Raman data set is associated with a second known sample and associated with one or more of: a second known disease state and a second known clinical outcome. A diagnosis is provided of whether the unknown sample has a first disease state or a second disease state by comparing the test Raman dataset to said plurality of reference Raman difference data sets in the reference Raman database using a chemometric technique.