Genetic algorithm-based wavelength selection in multicomponent spectrophotometric determinations by partial least square regression: application to a sulfamethoxazole and trimethoprim mixture in bovine milk

The simultaneous determination of sulfamethoxazole (SMX) and trimethoprim (TMP) mixtures in bovine milk by spectrophotometric method is, due to spectral interferences, a difficult problem in analytical chemistry. By means of multivariate calibration methods, such as partial least square (PLS) regression, it is possible to obtain a model adjusted to the concentration values of the mixtures used in the calibration range. A genetic algorithm (GA) is a suitable method for selecting the wavelengths for PLS calibration of mixtures with almost identical spectra without the loss of prediction capacity using a spectrophotometric method. In this study, a calibration model based on the absorption spectra in the 200–400 nm range for 25 different mixtures of SMX and TMP. Calibration matrices were formed from samples containing 0.25–20 and 0.3–21 μg mL-1 for SMX and TMP, at pH 10, respectively. The root mean squared error of deviation (RMSED) for SMX and TMP with PLS and genetic algorithm partial least square (GAPLS) were 0.242 and 0.066 μg mL-1, and 0.074 and 0.027 μg mL-1, respectively. This procedure allowed the simultaneous determination of SMX and TMP in synthetic and real samples and good reliability of the determination was proved.