Notice: Function _load_textdomain_just_in_time was called incorrectly. Translation loading for the pojo-accessibility domain was triggered too early. This is usually an indicator for some code in the plugin or theme running too early. Translations should be loaded at the init action or later. Please see Debugging in WordPress for more information. (This message was added in version 6.7.0.) in /home/doctoral/public_html/wp-includes/functions.php on line 6121
ICDSUPL3-F018 – University of Life Sciences in Lublin

ICDSUPL3-F018

Volume: 3, 2024
3rd International PhD Student’s Conference at the University of Life Sciences in Lublin, Poland:
ENVIRONMENT – PLANT – ANIMAL – PRODUCT

Abstract number: F018

DOI: https://doi.org/10.24326/ICDSUPL3.F018

Published online: 24 April 2024

ICDSUPL, 3, F018 (2024)


Comparison of multiple NIR instruments and machine learning algorithms for the detection of nitrogen-based adulteration in protein powders

Matyas Lukacs1, Flora Vitalis1, Zoltan Kovacs1*

1 Department of Measurement and Process Control, Institute of Food Science and Technology, Hungarian University of Agriculture and Life Sciences (MATE), Somlói 14-16, H-1118 Budapest, Hungary

* Corresponding author: kovacs.zoltan.food@uni-mate.hu

Abstract

Two spectroscopic devices, one benchtop and one handheld near-infrared spectrometer (NIRS), employing distinct signal processing techniques, were assessed for their efficacy in detecting pea protein powder adulteration at low-concentration levels. Adulteration of plant-based protein powders presents a prevalent issue in the food industry, largely due to the reliance on total nitrogen content determination as the primary analytical method for quality assurance. However, this method is inadequate in distinguishing between protein-derived nitrogen and nitrogen originating from other sources. To simulate adulteration scenarios, pea protein powder was blended with various compounds rich in nitrogen content, such as melamine, urea, taurine, and glycine, in the concentration range of 0.13–17.86%. Utilizing NIRS in conjunction with partial least squares (PLS) and artificial neural network (ANN) regression, predictive models were developed and compared to estimate foreign-compound concentrations even in the case of simultaneous adulteration. Models based on the data of both devices could reach comparably high accuracies with R2 values of up to 0.98, while models built with ANN could significantly outperform the PLSR ones as the matrix-complexity was gradually increased with introducing additional types of adulterants. Findings indicate the advantages of using algorithms capable of dealing with non-linear patterns derived from complex food matrices combined with NIRS in the quality control of protein powders.

Keywords: NIR spectroscopy, chemometrics, food fraud, portable device, machine learning


How to cite

M. Lukacs, F. Vitalis, Z. Kovacs, 2024. Comparison of multiple NIR instruments and machine learning algorithms for the detection of nitrogen-based adulteration in protein powders. In: 3rd International PhD Student’s Conference at the University of Life Sciences in Lublin, Poland: Environment – Plant – Animal – Product. https://doi.org/10.24326/ICDSUPL3.F018

Skip to content