HEURISTIC EVOLVING LATENT PROJECTIONS - RESOLVING 2-WAY MULTICOMPONENT DATA .2. DETECTION AND RESOLUTION OF MINOR CONSTITUENTS

The new method of heuristic evolving latent projections (HELP) is used for the resolution of two-component mixtures of isomers characterized by liquid chromatography with diode array detection. Four mixtures of drug isomers were analyzed with chromatographic resolutions of 0.2 and 0.5, and the minor component was present in 2 and 10%. Despite the very similar spectra of the isomers, the new method was able to resolve unambiguously all the mixtures into spectra and concentration profiles of the pure components. In comparison, the reference methods of overall factor analysis and evolving factor analysis failed for the mixtures with 2% relative concentration of the minor component. There are three major reasons for the good results obtained with the HELP method: (I) the HELP method enables the establishment of the "true" noise level in the data by local latent-variable analysis of the so-called zero-component chromatographic region, (II) the HELP method is capable of detecting and utilizing both spectral and chromatographic selective (one-component) information by a powerful combined visual/numeric approach, (III) the HELP method has the feature of a datascope. Thus, local selective spectral or chromatographic regions in the data are revealed In a stepwise inductive manner by focusing the analysis toward the most interesting chromatographic and spectral regions.