Solving matrix effect, spectral interferences and nonlinearity by generalized standard addition method coupled with radial basis functions-partial least squares: Application to simultaneous determination of drugs in urine

Abstract

We introduce here the coupling generalized standard addition method (GSAM) with radial basis functions-partial least squares (RBF-PLS) in inverse mode for analyte determination in the presence of matrix effect, spectral overlapping caused by interferences and nonlinearity. Therefore, in the procedure introduced, the whole sensor range can be used without afraid of collinearity problem encountered in application of GSAM with classical least squares and calibration can be made for each analyte, separately. The presence of nonlinearity was identified by a special case of principal components plots. Direct orthogonal signal correction (DOSC) was used as a preprocessing method. Resolution of binary and ternary mixtures of paracetamol (PAR), desxtromethorphan (DEX) and phenylephrine (PHE) in urine were successfully achieved without any sample pre-treatment and analyte separation. The mean recoveries for the real samples were between 98.2% and 107.4% in binary and between 95.9% and 100.3% in ternary mixtures. Relative standard deviations of the results were between 1.87% and 3.92% in binary and between 3.67 and 5.32 in ternary mixtures. Moreover, superiority of the proposed method over those which use classical least squares and partial least squares was shown.