NIRS calibration curves for the prediction of chemical compounds of green coffee Validación de curvas de calibración por NIRS para la predicción de compuestos químicos de café almendra

Near-infrared spectroscopy (NIRS) has been consolidated as a reliable, objective, reproducible, verifiable, inexpensive and low-environmental impact secondary analytical technique. This research aimed to carry out the validation of the equation developed from the NIRS technique for green coffee, which currently predicts 13 chemical compounds (caffeine, trigonelline, total chlorogenic acids, total lipids, fatty acids (palmitic, stearic, oleic, linoleic, linolenic and arachidic), isomers (3-CQA, 4-CQA and 5-CQA) and sucrose). The validation was carried out with 70 samples of Arabica green coffee (Coffea arabica L.), Castillo®, Cenicafé1 and Tabi varieties, produced in the departments of Cauca, Cesar and Caldas, which were analyzed to determine chemical compounds by international methods of analysis of the AOAC and standardized in Cenicafé; the samples were simultaneously analyzed in the NIRS equipment. The concordance between international analytical methods and NIRS was established from descriptive statistical analyses, Blan Altman concordance analysis and Pearson’s correlation between bias and magnitude. The relative error obtained by the NIRS technique for the compounds caffeine, sucrose, total chlorogenic acids, isomers of chlorogenic acids 4-CQA and 5-CQA, total lipids, arachidic, stearic and palmitic fatty acids was lower than 6.0%. The trigonelline compounds and linoleic acid showed an error of 7.0% and oleic acid 10.0%. The results confirm that with the calibration of the curve the NIRS technique can become a secondary analytical method.