Partial Least Squares Regression Method to Predict Docosahexaenoic and Eicosapentaenoic Acids in Fish Oil Supplements

Abstract

Fish oil dietary supplements have been linked to various health benefits due to the high concentration of omega-3 polyunsaturated fatty acid (ω-3 PUFA). The potential use of attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy with partial least squares regression (PLSR) was assessed to determine the docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA), and ω-3 PUFA in commercial fish oil capsules taking as a reference 1H NMR spectroscopy values. Comparing the results achieved by interval PLS (iPLS), synergy interval PLS (siPLS), and backward interval PLS (biPLS) algorithms, it was found that siPLS provided the best results. The proposed method predicted DHA with a coefficient of determination (R2) of 0.990, root mean square error of cross-validation (RMSECV) of 0.625%, and root mean square error of prediction (RMSEP) of 1.941. EPA (R2=0.976, RMSECV=0.789%, and RMSEP=2.795%) and ω-3 PUFA (R2=0.978, RMSECV=2.667%, and RMSEP=3.980%). The results indicated that ATR-FTIR and siPLS provided a robust method that could be employed in the analysis and quality control of fish oil supplement capsules. This method has the advantage of being simple, fast, and non-destructive for quantitative analysis.