GENETIC ALGORITHM-BASED METHOD FOR SELECTING WAVELENGTHS AND MODEL SIZE FOR USE WITH PARTIAL LEAST-SQUARES REGRESSION - APPLICATION TO NEAR-INFRARED SPECTROSCOPY

Genetic algorithms (GAs) are used to implement an automated wavelength selection procedure for use in budding multivariate calibration model s based on partial least-squares regression. The method also allows th e number of latent variables used in constructing the calibration mode ls to be optimized along with the selection of the wavelengths. The da ta used to test this methodology are derived from the determination of aqueous organic species by near-infrared spectroscopy. The three data sets employed focus on the determination of (1) methyl isobutyl keton e in water over the range of 1-160 ppm, (2) physiological levels of gl ucose in a phosphate buffer matrix containing bovine serum albumin and triacetin, and (3) glucose in a human serum matrix. These data sets f eature analyte signals near the limit of detection and the presence of significant spectral interferences. Studies are performed to characte rize the signal and noise characteristics of the spectral data, and op timal configurations for the GA are found for each data set through ex perimental design techniques. Despite the complexity of the spectral d ata, the GA procedure is found to perform web, leading to calibration models that significantly outperform those based on full spectrum anal yses. In addition, a significant reduction in the number of spectral p oints required to build the models is realized.