Chemometric analysis of mass spectral and spectroscopic data for improved NPS identification

The field of forensic drug analysis has evolved substantially over the past several decades. Increases in new substances present a notable analytical challenge because typical forensic methodology (Gas Chromatography-Mass Spectrometry [GC-MS], for instance) is frequently incapable of identifying the small structural differences between similar compounds. Positional isomers, wherein the only difference is the precise location of a small moiety on an aromatic ring, often produce visually indistinguishable mass spectra. In some instances, these isomers also display similar chromatographic behavior, further complicating the analysis. The work presented in this thesis offers an alternative method of handling challenging drug exhibits by focusing on advanced data science techniques rather than improvements in instrumental hardware. The application of chemometric methods to data that is commonly generated in the regular course of forensic analysis is likely to be more accessible to laboratories than the purchase of additional instruments or by conducting supplementary time-consuming chemical analysis. The instrumental techniques surveyed in this thesis include GC-MS, Direct Analysis in Real Time – Time-of-Flight Mass Spectrometry (DART-TOF MS) and Gas Chromatography – Infrared Spectroscopy (GC-IR). The chemometric methods utilized include Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), the sequential combination of the two methods (PCA-LDA), Welch’s t-test, and the Random Forest algorithm. For each instrumental method, chemometrics provided increased objectivity and improved the discrimination capabilities when faced with the challenge of distinguishing between highly similar compounds.